WATER-LUBRICATED OIL FLOWS IN A COUETTE APPARATUS

A Thesis Submitted to the College of Graduate Studies and Research in Partial Fulfillment of the Requirements for the Degree of Masters of Science in the Department of Chemical Engineering University of Saskatchewan, Saskatoon

By

Rasheed 0 luseun Bello

©Copyright Rasheed Oluseun Bello, January 2004. All rights reserved

The University of Saskatchewan claims copyright in conjunction with the author. Use shall not be made of the material contained herein without proper acknowledgement. In presenting this thesis in partial fulfillment of the requirements for a Postgraduate degree from the University of Saskatchewan, I agree that the Libraries of this University may make it freely available for inspection. I further agree that permission for copying of this thesis in any manner, in whole or in part, for scholarly purposes may be granted by the professor or professors who supervised my thesis work or, in their absence, by the Head of the Department or the Dean of the College in which my thesis work was done. It is understood that any copying or publication or use of this thesis or parts thereof for financial gain shall not be allowed without my written permission. It is also understood that due recognition shall be given to me and to the University of Saskatchewan in any scholarly use which may be made of any material in my thesis.

Requests for permission to copy or to make other use of material in this thesis in whole or part should be addressed to:

The Head of the Department Department of Chemical Engineering University of Saskatchewan Saskatoon, Saskatchewan Canada S7N 5C9.

1 ABSTRACT

Self-lubricating flow of bitumen froth occurs when water droplets dispersed in the viscous oil phase form a lubricating layer surrounding a bitumen-rich core in a commercial pipeline operated by Syncrude Canada Ltd. In this study, self-lubricating flow was modelled experimentally by shearing water-in-oil emulsions in a Couette cell device (viscometer). Bitumen and two commercially available lube oils (N-Brightstock and Shellflex 81 0) were tested. Dispersed phase water concentrations ranged from 10 to 35 wt%. The ability to produce self-lubricating flow was a function of dispersed phase water concentration, temperature, oil viscosity, spindle speed and spindle size (gap width). For tests conducted at constant temperature, the critical spindle speed required to achieve self-lubricating flow was found to decrease with increasing water concentration. The critical spindle speed at which self-lubricating flow was achieved was found to increase with increasing temperature for the emulsions tested. Flow maps showing the critical spindle speed as a function of temperature indicate a clearly delineated transition region between viscous (non-lubricating) and self-lubricating flow. The critical spindle speed required to achieve self-lubricating flow was found to be similar when the viscosities of different oils were matched by varying the operating temperature. Self-lubricating flow was achieved with small and medium diameter spindles but not with the large diameter spindle. Models of self-lubricating flow in the Couette cell were developed to predict the thickness of the water layer that forms during self-lubricating flow. The thickness of the water layer calculated from these models is less than that determined by others for pipeline flows. The results of this study indicate that the ability to produce and maintain self­ lubricated flow is highly dependent upon water concentration, emulsion temperature and the continuous phase viscosity. Further studies of these model emulsions using pipeline flows are required.

11 ACKNOWLEDGEMENTS

I want to thank Almighty Allah for making this project a reality and for providing me with the strength to persevere throughout the duration of my Masters.

I would like to thank Syncrude Canada Ltd., the University of Saskatchewan and the Petroleum Society of the Canadian Institute of Mining and Metallurgy for financial support throughout my Masters program.

I would like to thank the following people:

My Supervisors, Dr. R. Sumner and DrS. Sanders for their immense contributions and assistance towards completing this project. I want to thank them for providing me with the opportunity to work on a groundbreaking project.

The Members of my Thesis Committee; Pro£ C. Shook, Dr. M. McKibben, and Prof. G. Hill; for their contributions and recommendations towards making this a better project

Dr. R. Gillies and the entire staff of the Saskatchewan Research Council Pipeflow Centre for providing technical assistance for my project and making my time at the SRC comfortable.

Alexis McPherson, Claire Sirois and Andrew May for their assistance and contributions towards making this research a reality.

My Parents; Prof. R.A.Bello and Mrs M.A. Bello; for helping to make me what I am today and for their continuous prayers and patience throughout my project. My siblings Biodun, Bidemi and Tolu for their support and prayers.

111 TABLE OF CONTENTS

PERMISSION TO USE i ABSTRACT li ACKNOWLEDGEMENTS ill TABLE OF CONTENTS iv LIST OF TABLES vi LIST OF FIGURES vili NOMENCLATURE xili

1 INTRODUCTION 1 1.1 Background 1 1.2 Project Definition 4 1.3 Research Objectives 5

2 LITERATURE REVIEW 6 2.1 Introduction 6 2.2 Couette Flow In A Concentric Cylinder System (Viscometer) 7 2.3 Emulsion Rheology 9 2.4 Oil-Water Pipeline Flows 10 2.4. 1 Core-Annular Flow 13 2.4.1.1 Water-Assisted Flow 16 2.4.1.2 Self-Lubricating Flow 16 2.5 Particle and Droplet Migration in Viscous Fluids 26

3 EXPERIMENTAL EQUIPMENT AND PROCEDURES 31 3 .1 Introduction 31 3.2 Characterization of Oils 32 3.2.1 Density Determination 32 3.2.2 Viscosity Determination 33 3.3 Emulsion Preparation 34 3.4 Droplet Size Analysis 36 3.5 Couette Flow Test Equipment 37 3.5 .1 Haake Rotovisco Viscometer 3 8 3.5.2 Haake Rheostress Rheometer 39

4 RESULTS AND DISCUSSION 42 4.1 Continuous Phase (Oil) Characterization 42 4.1.1 Density Measurements 42 4.1.2 Viscosity Measurements 46

IV 4.1.2.1 Effect ofTemperature Gradient 49 4.1.2.2 Effect ofViscous Heating 51 4.1.2.3 Verification ofViscometer Accuracy 53 4.1.2.4 Rheology of Oils 53 4.2 Emulsion Preparation and Characterization 58 4.2.1 Emulsion Stability 59 4.2.2 Assessment of Droplet Size Distribution 64 Reproducibility and Variability 4.3 Self-Lubricating Flow in a Couette Cell 68 4.3 .1 Development of a Criterion for Assigning 69 Self-Lubricating Flow 4.3.2 Reproducibility of Experimental Runs 74 4.3.3 Comparison of Observed Initial Torques with 76 Theoretical Predictions 4.4 Development of Self-Lubricating Flow Maps 77 4.4.1 N-Brightstock Emulsions 77 4.4.2 Other Emulsions 83 4.5 Effect of Shear on Water Droplet Size 84 4.6 Effect of Water Concentration on Self-Lubricating Flow 90 4.7 Effect of Shear History on Self-Lubricating Flow 93 4.8 Effect of Spindle Size (Gap Width) on Self-Lubricating Flow 95 4.9 Effect of Type of Oil on Self-Lubricating Flow 97 4.10 Visual Observations 100 4.11 Summary 107

5 MODEL DEVELOPMENT FOR SELF-LUBRICATING 110 FLOW IN A COUETTE CELL 5.1 Model I - Shearing in Water Layer Only 111 5.2 Model II - Shearing Throughout Couette Cell Gap 114 5.3 Comparison of Experimental Results with Model Predictions 118

6 CONCLUSIONS 120

7 RECOMMENDATIONS 123

8 REFERENCES 124

APPENDICES Appendix A - Oil Density and Viscosity Data 130 Appendix B - Couette Flow Experimental Data 139 Appendix C - Derivation of the Viscous Heating Model 171 Appendix D - Derivation of the Models for Self-Lubricating 173 Flow in a Couette Cell Appendix E - Derivation of the Equations for Single Phase 179 Couette Flow in a Viscometer

v LIST OF TABLES

TABLE TITLE PAGE

3.1 Haake RV3 Viscometer Specifications 38 3.2 Haake RS 150 Rheometer Specifications 40 4.1 Effect of Spindle Size on Viscous Heating of 52 N-Brightstock oil

4.2 Mean Droplet Diameter and Standard Deviation 61 Data for 25 wt% water-in-N-Brightstock Oil Emulsion Stability Tests

4.3 Mean Droplet Diameter and Standard Deviation 66 Data for 30 wt% water-in-N-Brightstock Oil Emulsion Reproducibility Tests

4.4 Summary of Experimental Parameters Studied 68 During the Investigation of Self-Lubricating Flow in A Couette Cell

4.5 Measure of Reproducibility of Couette Flow 75 Experimental Results for 30 wt% water-in-N- Brightstock Oil Emulsion

4.6 Comparison of Initial and Final Torques Obtained 76 from Couette Flow Experiments with 30 wt% water-in- N-Brightstock Oil Emulsion at 30 and 50°C

4.7 Comparison of Observed Initial Torques Obtained for 77 Self-Lubricating Flow of 30 wto/o water-in-N-Brightstock Oil Emulsion with Theoretical Predictions

5.1 Comparison of Water Layer Thicknesses Using 118 Models I and II

A.l Measured N-Brightstock Oil Density as a Function 130 of Temperature

A.2 Measured Shell flex 810 Oil Density as a Function of 131 Temperature

Vl A.3 Measured N-Brightstock Oil Viscosity as a Function 131 of Temperature

A.4 Measured Shellflex 81 0 Oil Viscosity as a Function 132 of Temperature

B.1 Couette Flow Experimental Results for N-Brightstock, 140 Shellflex 810 and Bitumen Emulsions

B.2 Reproducibility of Couette Flow Experimental Results 155 Obtained with 30 wt% water-in-N-Brightstock Oil Emulsion

B.3 Couette Flow Experimental Results for 30 wt% water 157 -in-N-Brightstock Oil Emulsion

B.4 Couette Flow Experimental Data for Bitumen Froth 159 B.5 Effect ofWater Concentration on Self-Lubricating 160 Flow

B.6 Effect of Initial Torque Reduction Time on Self- 161 Lubricating Flow of 30 wt% water-in-N-Brightstock Oil Emulsion

B.7 Couette Flow Data for 30 wt% water-in-N-Brightstock 162 Oil Emulsion Sheared with Z31 Spindle (Shear Rate, Temperature as Independent Variables)

B.8 Couette flow data for 30 wt% water-in-N-Brightstock 164 Oil Emulsion Sheared with Z38 Spindle (Shear Rate, Temperature as Independent Variables) B.9 Couette Flow Data for 30 wt% water-in-N-Brightstock 165 Oil Emulsion Sheared with Z41 Spindle (Shear Rate, Temperature as Independent Variables)

B.10 Effect of Oil Viscosity on Self-Lubricating Flow 166

vii LIST OF FIGURES

FIGURE TITLE PAGE

1.0 Schematic Illustration of a Truck-and-Shovel Mining, 2 Slurry Preparation and Oil Sand Extraction Operation

2.1 Couette Flow of a Single Phase, Incompressible 8 Newtonian Fluid in a Concentric Cylinder Viscometer

2.2 Shear Stress as a Function of Time for a Newtonian 10 Fluid at Constant Temperature and Shear Rate

2.3 Horizontal Pipeline Flow Patterns of Oil and Water 11 (based on Charles et al., 1961)

2.4 Stratified Flow Pattern in Horizontal Pipeline Flow 12 of Oil and Water (based on Russell et al., 1959)

2.5 Wavy Core-Annular Flow of Oil and Water 13 (based on Bannwart, 2001)

2.6 Measured Pipeline Hydraulic Gradient for Bitumen Froth 19 as a Function of Time (based on Neiman, 1986)

2.7 Measured Torque of Bitumen Froth as a Function of Time 20 in a Concentric Cylinder Viscometer (Sumner, 2003)

3.1 Schematic Diagram of the Greerco Homomixer Model 1L 35 4.1 Measured N-Brightstock Oil Density as a Function of 43 Temperature (Temperature Range of 10 to 50°C)

4.2 Measured Shell flex 81 0 Oil Density as a Function of 45 Temperature (Temperature Range of 10 to 50°C)

4.3 Newtonian Model for Measured Torque and Angular 47 Velocity Data ofN-Brightstock Obtained Using Z38 Ti Spindle at a Temperature of 50°C

Vlll 4.4 Newtonian Model Fit for Measured Torque and Angular 48 Velocity Data ofN-Brightstock Oil Obtained Using Z31 Ti Spindle at a Temperature of 50°C

4.5 Measured Torque Data ofN-Brightstock Oil as a Function 50 of Time Using Z31 Ti Spindle at a Constant Spindle Speed of 50 rpm and a Temperature of 50°C

4.6 Viscosity Measurements ofN-Brightstock Oil as a Function 54 of Temperature {Temperature Range of22.5 to 50°C)

4.7 Viscosity Measurements of Shell flex 81 0 Oil as a Function 57 of Temperature (Temperature Range ofO to 50°C)

4.8 Comparison of the Droplet Size Distributions of 25 wto/o water 60 in-N-Brightstock Oil Emulsion Samples Band F

4.9 Comparison of Droplet Size Distributions of30 wt% water- 63 in-N-Brightstock Oil Emulsion Sampled Immediately (0 day) and After 24 hrs

4.10 Droplet Size Distribution of30 wt% water-in-N-Brightstock 67 Oil Emulsion. Emulsion 1,Sample 1a, Emulsion 2, Sample 2a

4.11 Self-Lubricating Flow of30 wt% water-in-N-Brightstock Oil 70 Emulsion

4.12 Marginal Self-Lubricating Flow of30 wt% water-in-N- 71 Brightstock Oil Emulsion

4.13 Viscous (Non-Lubricating Flow) of30 wt% water-in-N- 72 Brightstock Oil Emulsion

4.14 Flow Map for 30 wt% water-in-N-Brightstock Oil Emulsion 78 (Temperature and Spindle Speed as Independent Variables)

4.15 Flow Map for 30 wt% water-in-N-Brightstock Oil Emulsion 80 (Measured Torque and Spindle Speed as Independent Variables)

4.16 Flow Map for water-in-Bitumen Emulsion (Temperature and 82 Spindle Speed as Independent Variables)

4.17 Flow Map for 30 wt% water-in-Shellflex 810 Oil Emulsion 84 (Temperature and Spindle Speed as Independent Variables)

ix 4.18a Photomicrographs Showing Water Droplets in Oil Matrix 86 for Self-Lubricating Flow Case Before Shear

4.18b Photomicrographs Showing Water Droplets in Oil Matrix 86 for Self-Lubricating Flow Case After Shear

4.19 Droplet Size Distribution for 30 wt% water-in-N- 87 Brightstock Oil Emulsion Before and After Shearing With Concentric Cylinder Viscometer for Self-Lubricating Flow

4.20a Photomicrographs Showing Water Droplets in Oil Matrix 89 After Shear for Non-Lubricating Flow Case Before Shear

4.20b Photomicrographs Showing Water Droplets in Oil Matrix 89 After Shear for Non-Lubricating Flow Case After Shear

4.21 Droplet Size Distribution for 30 wt% water-in-N- 90 Brightstock Oil Emulsion Before and After Shearing with Concentric Cylinder for Non-Lubricating Flow

4.22 N-Brightstock Oil Emulsion Water Concentration as a 92 Function of Critical Spindle Speed Required to Achieve Self-Lubricating Flow with Temperature as a Parameter

4.23 Schematic Representation of Torque Reduction Time for 93 Self-Lubricating Flow

4.24 Initial Torque Reduction Time as a Function of Spindle 94 Speed with Temperature as a Parameter for Self-Lubricating Flow of 30 wt% water-in-N-Brightstock Oil Emulsions

4.25 Effect of Spindle Size on the Self-Lubricating Flow of 96 30 wt% water-in-N-Brightstock Oil Emulsion

4.26 Oil Viscosity as a Function of Critical Spindle Speed 99 Required to Achieve Self-Lubricating Flow of 30 wt% water-in-Oil Emulsion

4.27 Schematic Diagram (Plan View) of Progression from 101 Original Emulsion to Non-Lubricating Flow

4.28 Schematic Diagram (Plan View) of Progression from 103 Original Emulsion to Self-Lubricating Flow

X 4.29 Schematic Diagram of the Flow Structure ofN-Brightstock 104 Oil Emulsion during Self-Lubricating Flow

5.1 Schematic Illustration of Model I (Shearing in Water 111 Layer Only)

5.2 Schematic Illustration of Model II (Shearing in Water 114 and Emulsion Layers)

5.3 Predicted Velocity Profile for Self-Lubricating Flow of 117 an Emulsion in a Couette Cell (Proceeding from Spindle towards Cup) from Model II

A.1 Newtonian Model Fit for Measured Torque and Angular 133 Velocity Data of Cannon S600 Viscosity Standard Obtained Using Z31 Ti Spindle (15.725 mm Radius) at a Temperature of20°C

A.2 Bingham Model Fit for Measured Torque and Angular 134 Velocity Data of Cannon S600 Viscosity Standard Obtained Using Z31 Ti Spindle (15.725 mm Radius) at a Temperature of20°C

A.3 Natural Logarithm of the Measured Newtonian Viscosities 135 ofN-Brightstock Oil as a Function of the Inverse of the Absolute Temperature (Based on Equation 4.5)

A.4 Natural Logarithm of the Measured Newtonian Viscosities 136 of Shellflex 810 Oil as a Function of the Inverse of the Absolute Temperature (Based on Equation 4.5)

A.5 Newtonian Model Fit for Measured Torque and Angular 137 Velocity Data for Shell flex 810 Oil Obtained Using Z31 Ti Spindle (15.725 mm Radius) at a Temperature of50°C

A.6 Newtonian Model Fit for Measured Torque and Angular 138 Velocity Data for Coker-Feed Bitumen Obtained Using Z31 Ti Spindle (15.725 mm Radius) at a Temperature of50°C

B.1 Flow Map for 10 wt% water-in-N Brightstock Oil Emulsion 167 (Temperature and Spindle Speed as Independent Variables)

B.2 Flow Map for 17 wt% water-in-N-Brightstock Oil Emulsion 168 (Temperature and Spindle Speed as Independent Variables)

Xl B.3 Flow Map for 25 wt% water-in-N-Brightstock Oil Emulsion 169 (Temperature and Spindle Speed as Independent Variables)

B.4 Flow Map for 35 wt% water-in-N-Brightstock Oil Emulsion 170 (Temperature and Spindle Speed as Independent Variables)

D.l Schematic Illustration of Model II (Shearing in Water and 174 Emulsion Layers)

E.l Couette Flow of a Single Phase, Incompressible Newtonian 179 Fluid in a Concentric Cylinder Viscometer

E.2 Couette Flow Velocity Profile in a Concentric Cylinder 185 Viscometer as a Function of Spindle Size and Radial Position Progressing from Spindle towards Cup

E.3 Couette Flow Shear Rate Profile in a Concentric Cylinder 186 Viscometer as a Function of Spindle Size and Radial Position Progressing from Spindle towards Cup

Xll NOMENCLATURE

A constant for Andrade equation (Equation 4.4) A emulsion layer (Chapter 5) b constant for Andrade equation (Equation 4.4) B water layer (Chapter 5) C emulsion layer (Chapter 5) Cp specific heat capacity Ct,C2 constants for velocity profile (Chapter 5) dN,L number, length mean diameter (arithmetic mean diameter) ds,v surface, volume mean diameter (Sauter mean diameter) D pipe diameter Dt, D2 constants for velocity profile (Chapter 5) Et,E2 constants for velocity profile (Chapter 5) g acceleration due to gravity L spindle length n spindle speed r radial distance r, e, z cylindrical coordinates

R~,R2 viscometer spindle radius, cup radius (Chapter 5) Rt viscometer spindle radius

R2,R3 water layer bounding radii (Chapter 5)

~ cup radius t duration of shear (Equation 4.3) T temperature T torque exerted on spindle v local velocity in Couette flow y shear rate J.1 fluid viscosity f.lp plastic viscosity

Xlll J.lA emulsion layer viscosity (Chapter 5) J.lB water layer viscosity (Chapter 5) J.lc emulsion layer viscosity (Chapter 5) p fluid density

't shear stress

'ty yield stress co angular velocity of spindle

XIV Chapter One

INTRODUCTION

1.1 Background Diminishing reserves of conventional crude oil have resulted in increased production from alternative sources of crude oil such as the Athabasca oil sands deposit of Northern Alberta. Oil sands represent a source of bitumen that can be upgraded or converted into the lighter fractions of crude oil (naphtha and gas oils). According to the World Energy Council (2003), the Athabasca oil sands deposit is estimated to contain about 1. 7 trillion barrels of bitumen. The recoverable reserves in the Athabasca Oil sands deposit are estimated to be about 300 billion barrels. This is more than Saudi Arabia's reserves of 262 billion barrels of conventional crude oil (Western Oil Sands, 2003).

The Athabasca oil sands typically contain about 10 to 12% bitumen, 80 to 85% mineral matter including clay and sand, and 4 to 6% water by weight (Government of Alberta, 2003). Bitumen is a very viscous substance. The viscosity of bitumen-in-place at reservoir temperature in the Athabasca deposit is approximately 1,000 Pa· s (Schramm and Kwak, 1988). This makes bitumen immobile and renders its production difficult. The immobility of bitumen distinguishes it from heavy oils which are more mobile at reservoir temperatures (Chilingarian and Yen, 1978).

Different techniques are used to produce bitumen from the oil sands deposits, but these can broadly be categorized as either in-situ production or surface-mining and extraction methods. In-situ production methods include well-bore steam ~jection, the SAGO

1 Process (Steam-Assisted Gravity Drainage), or the injection of solvents to reduce the in­ situ bitumen viscosity. The produced fluid is typically a water-in-oil emulsion. Related techniques are also used to exploit heavy oil (as opposed to oil sands) reserves.

Where oil sands deposits are located closer to the surface, overburden is stripped and oil sand is mined from large, open pits. The ore is mined with large shovels and transported to a comminution and sizing plant by large trucks. The oil sand is then mixed with water, and transported via pipeline as a dense slurry from the mine to the extraction plant. Variations of the Clark Hot Water Process (Shaw et al., 1996) are used by the three current oil sands mine operators (Suncor Energy Inc., Syncrude Canada Ltd. and Albian Sands) to separate the bitumen from the other oil sand constituents. Figure 1.0 shows a typical water-based extraction process flow sheet (Sanders et al., 2002).

f) SluJTYing

Oil Sand Slurry

Oil Sand Hydro transport Underflow ~Bitumen Separation

Figure 1.0 Schematic illustration of a truck-and-shovel mining, slurry preparation and oil sand extraction operation (after Sanders et al., 2002).

2 The intermediate product of the gravity separation process used to extract bitumen is referred to as bitumen froth. It is unlike the more conventional types of 'froth' that are found in the mineral processing industry, except that it does contain a significant quantity of air (as much as 50% by volume). Bitumen froth is a complex, multi-phase mixture that contains, in addition to air, approximately 60% bitumen (by mass), 30% water and 10% solids. It is generally considered to be a bitumen-continuous mixture and thus is characterized as having a very high viscosity, of the order of 20 Pa·s at 50°C (Neiman, 1986).

The challenges of transporting this viscous mixture, whether through short, intra-site transfer lines or long-distance, inter-site pipelines, are very similar to those faced by heavy oil producers everywhere: pumping viscous fluids requires significant energy requirements (Charles, 1960; Nunez et al., 1998). In the heavy oil industry, different methods have been employed to reduce crude viscosity and thus reduce the energy requirements for transport. These methods include heating, the addition of a hydrocarbon diluent, emulsification or partial upgrading (Nunez et al., 1998). Significant capital investment and operating costs are associated with these methods.

A technique referred to as lubricated pipelining (Charles, 1960; Sanders et al., 2004) or core-annular flow (Joseph, 1997; Nunez et al., 1998) has been studied extensively and has been used successfully in some industrial applications. The process used by Syncrude Canada Ltd. to transport bitumen froth from its remote extraction facilities to a centralized treatment and upgrading facility is known as Self-Lubricating Flow, or Natural Froth Lubricity (Neiman et al., 1999; Joseph et al., 1999). Self-lubricating flow involves the migration of some of the originally dispersed water droplets to a region of high shear near the pipe wall (K.ruka, 1977). These water droplets subsequently coalesce to form an annular layer around a bitumen-rich core. The presence of this water layer results in a greatly reduced flow resistance, thereby reducing pumping energy requirements. The measured pressure gradients for self-lubricated flow have been observed to be two orders of magnitude lower than would be predicted based on the viscosity ofbitumen froth (Neiman, 1986; Shook et al., 2002).

3 Syncrude's bitumen froth pipeline is the largest lubricated pipelining installation in the world. It is 35 km in length and 0.91 min diameter, and is designed to transport more than 26 x 106 m3 bitumen froth annually (Mankowski et al., 1999). Because the pipeline's capacity is so great, and its reliability so critical, Syncrude and its external research providers continue to look for ways to expand the pipeline's operating envelope. Additionally, Syncrude researchers continue to refine their models used to predict the variation of pipeline friction losses as a function of key operating parameters, such as froth composition, flow rate and temperature. The work conducted as part of this thesis focuses primarily on the relationship between shear, continuous phase viscosity and dispersed phase (water) concentration and the ability to produce self-lubricating flow.

1.2 Project Defmition Numerous pipe flow tests ofbitumen froth have been conducted (Neiman, 1986; Joseph et al., 1999; Shook et al., 2002; Sumner et al., 2003; Sanders et al., 2004) along with some experimental studies using other shearing devices, such as Couette flow cells (Sumner et al., 2003; Sanders et al., 2004). One of the difficulties that arises during these tests is the variability of the froth composition. Because froth samples are taken during the processing of many different oil sand ores, their composition can vary in terms of water content, water droplet size, water chemistry, fine solids content and coarse solids content. Thus it would be advantageous to develop an idealized froth analogue, where water is emulsified in a well characterized viscous oil (Sumner et al., 2003).

Additionally, pipe flow tests of bitumen froth require significant sample volumes (Sumner et al., 2003) and thus it is prohibitive to conduct parametric studies in this manner. Pipe flow tests are more suited to the evaluation of bulk flow properties, such as the effect of temperature and flow rate on frictional pressure gradients, but are less appropriate for the study of the effects of composition changes.

4 Limited research in this area (Sumner et al., 2003) has shown that it is possible to produce a water-in-oil emulsion that will provide self-lubricating flow in a Couette cell. This study builds on the previous work of Sumner et al. (2003).

1.3 Research Objectives The overall objective of this research project is to further the understanding of phenomena governing self-lubricated flow of bitumen froth. More specifically, the objectives of this study are:

• To determine if bitumen froth and model water-in-oil emulsions exhibit similar behaviour (flow characteristics) when sheared in a Couette flow cell. • To develop a technique that produces a stable emulsion. • To study and appropriately characterize different types of viscous oils that could serve as bitumen analogues for self-lubricating flow studies. • To evaluate the consistency and reproducibility of self-lubricating flow studies performed with a Couette flow cell. • To develop flow maps that show the conditions for which self-lubricating flow will (or will not) occur. • To show how different parameters (such as water concentration, spindle speed, shear profile, temperature and oil properties) affect the formation and stability of self-lubricated flow in a Couette flow cell. • To propose mathematical models which represent self-lubricated flow in a Couette flow device.

5 CHAPTER TWO

LITERATURE REVIEW

2.1 Introduction

Heavy viscous oils may be transported by pipeline. However, the high viscosity of the oils results in significant pressure gradients and thus considerable pumping energy requirements (Charles, 1960). Different methods have been employed to reduce the high oil viscosity to enable reduced flow resistance. These methods include the addition of a hydrocarbon diluent and heating of the pipeline at short intervals along its length. These two methods can be inconvenient and costly. The use of diluents requires that light oil be available in the production area and heating the pipeline will lead to increased operating costs (Oliemans and Ooms, 1986). Emulsification of crude oil in water using a surfactant is mentioned by McKibben et al. (2000b) as another method of transporting heavy oil.

An alternative method of transporting heavy oil is by core-annular flow. Core-annular flow is such that the viscous oil is at the core of the pipe and is surrounded by a lubricating water annulus. This method has been shown to result in low pipeline pressure gradients (Charles et al., 1961). This method involves the addition of water to viscous oil (Water-Assisted Flow) or subjecting a water-in-oil emulsion to significant shear (Self-Lubricating Flow).

Kruka (1977) obtained self-lubricating flow of a water-in-oil emulsion by shearing it in a pipe. The shear causes the migration of the water droplets in the oil continuum to regions of high shear near the pipe wall. These water droplets eventually coalesce to form an annular water layer around the viscous oil in the core of the pipe.

6 Sumner et al. (2003) and Sanders et al. (2004) employed a concentric cylinder device as a means of shear to cause the self-lubricating flow of bitumen froth. Migration and coalescence of water droplets described by Kruka (1977) was also believed to occur in the experiments by Sumner et al. (2003) and Sanders et al. (2004) as indicated by a substantial reduction in measured torques to a persistent low value which is associated with a lubricating water layer.

A concentric cylinder viscometer will be employed in this research to shear water-in­ model oil emulsions as a continuation of work by Sumner et al. (2003).

The fundamental principles involved in single phase Couette flow in a concentric cylinder viscometer will be reviewed in Sections 2.2. Emulsion rheology will be reviewed in Section 2.3, to enable better understanding of emulsion behaviour under shear. Oil and water pipeline flow patterns will be reviewed in Section 2.4. Core­ annular flow and self-lubricating flow will be reviewed in Sections 2.4.1. Particle and droplet migration will be reviewed in Section 2.5 with the objective of understanding the possible mechanisms responsible for the migration of water droplets.

2.2 Couette Flow In A Concentric Cylinder System

A concentric cylinder system is illustrated in Figure 2.1. The fluid is placed in a

stationary cup of radius R2 and is sheared across the gap between the spindle and cup by

a spindle of radius R1 (also known as a bob or rotor) rotating at an angular velocity ro.

7 A large annular gap between the spindle and the cup will be assumed. Steady state and laminar flow will be assumed. The fluid will also be assumed to be incompressible and Newtonian. A detailed derivation is provided in Appendix E

The velocity profile as illustrated in Figure 2.1 is such that the velocity is a maximum at the spindle and decreases to zero at the cup. This is illustrated with an example in Appendix E.

r

Cup

Spindle

Figure 2.1 Couette flow of a single phase, incompressible Newtonian fluid in a concentric cylinder viscometer.

The relationship between the torque exerted by the fluid on the spindle and the rotational angular velocity of the spindle is given by Equation E.13 (Appendix E) as

8 (2.1)

with spindle angular velocity (radls), 21ln OJ=- 60 The spindle speed, n, has units of revolutions per minute (rpm).

2.2 Emulsion Rheology An emulsion contains two immiscible phases, one of which is dispersed in the other in the form of droplets of microscopic or submicroscopic size (Sherman, 1988). Khan (1996) indicates that in an emulsion, free energy is reduced through droplet coalescence. A surfactant is normally introduced into the emulsion to reduce the coalescence rate and prolong the emulsion state.

In some cases, the general rheological behaviour of single phase fluids is applicable to emulsions. Fluids can be classified according to their rheological behaviour as Newtonian or non-Newtonian. Newtonian fluids obey the following linear relation at constant temperature and pressure:

'= p,y (2.2) The shear stress in the fluid is proportional to the shear rate applied to the fluid. The proportionality constant is the viscosity of the fluid.

The relationship between shear stress and time for a Newtonian fluid (at constant shear rate and temperature) is shown in Figure 2.2.

9 t or

time Figure 2.2 Shear stress as a function of time for a Newtonian fluid at constant temperature and shear rate.

It can be observed from Figure 2.2 that the shear stress or viscosity is constant with time. Thus, in the case of the concentric cylinder viscometer geometry considered earlier, a plot of Torque (which is a function of the shear stress) against time for a Newtonian fluid should yield the same plot as shown above at a constant shear rate. This observation is important for the emulsion work in this research and will be referred to subsequently.

2.4 Oil-Water Pipeline Flows Charles et al. ( 1961) were the first to provide details on the different flow patterns present in oil-water flows. Oil-water flow experiments were conducted by Charles et al. (1961) with three different oils (viscosities of 6.29, 16.8 and 65 mPa· s) in a 26 mm 3 pipe. The density of these oils was made equal to that of water (998 kglm ) with the aid of an additive. It was discovered that oil and water form different flow patterns

10 depending on their flowrates (velocity). As the oil flowrate was decreased at a constant water flowrate, the flow pattern changed from one where water droplets are suspended in oil to a continuous water layer formed around an oil core (concentric oil in water) and to an oil slug flow. At high water flowrates, a dispersed oil in water flow pattern was obtained. These flow patterns obtained by Charles et al. (1961) are illustrated in Figure 2.3. The flow regimes for the three oils were similar except at low water velocity and low water-oil ratios where the high viscosity oil (65 mPa· s) exhibited a dispersed water­ in-oil flow and water slug flow. This was attributed to the preferential wetting of the pipe wall by this oil at low water flowrates. This high viscosity oil was prepared from a lubricating oil and the observed flow patterns appear to agree with the water slugs in oil flow pattern observed by McKibben et al. (2000a) with lubricating oils at low velocities.

Water droplets in oil Concentric oil in water (Core-Annular Flow) •••• Oil slugs in water Oil droplets dispersed in water

Oil Water

Figure 2.3 Horizontal pipeline flow patterns of oil and water (based on Charles et al., 1961).

11 Stratified and dispersed oil-in-water flow patterns were observed by Russell et al. (1959) in their experiments with a mineral oil (viscosity= 18 mPa· s, density= 834 3 kglm ) in a 25.4 nun pipe. This is illustrated in Figure 2.4

Oil

Water

Stratified flow

Figure 2.4 Stratified flow pattern in horizontal pipeline flow of oil and water (based on Russell et al., 1959).

McKibben et al. (2000a) also determined that a flow pattern with a water slug enveloping oil occurred with heavy oils at low velocities. This flow pattern is such that water slugs exist in a continuous oil phase and contain oil. Ooms et al. (1984) discovered in their experiments that below a certain critical velocity, a stratified flow pattern was obtained, with oil in the upper part of the pipe and water in the lower part of the pipe.

Charles (1960) noted that pipeline pressure gradients measured with heavy oil-water mixtures were an order of magnitude less than the gradients measured with heavy oil flowing alone at the same flow rate. Observations of the flow patterns could not be made because the oil was discovered to wet the pipe. The flow pattern was presumed to be stratified flow based on comparisons with results by Russell et al. ( 1959). Results obtained by McKibben et al. (2000a) suggest that the flow regime might have been water slugs enveloping oil which exhibits a reduction in time-averaged pressure gradients.

12 Charles et al. (1961) found that the largest pressure gradient reductions occurred with the concentric oil-in-water, and oil slugs in water flow patterns. Pressure gradient reduction factors were found to increase with the viscosity of the oils. The concentric oil-in-water flow pattern is more commonly called core-annular flow and is of great benefit in the industrial pipeline transport of heavy oil. It is also of importance to this study.

2.4.1 Core-Annular Flow Core-annular flow belongs to a general class of flows called lubricated flows. Lubricated flow of two phases involves conditions in which there is a separation of the phases with one phase located in the high shear region near the pipe wall in a pipe and the other phase at the core of the pipe. Lubricated flow is not limited to two liquids alone. Other examples include turbulent gas lubricated by water, slurries lubricated by water and turbulent gas lubricated by a concentrated dispersion of solids (Joseph, 1998). In core-annular flow, the oil is at the core and is surrounded by a water annulus.

Joseph et al. (1997) describe two types of core-annular flow: Perfect Core Annular Flow and Wavy Core Annular Flow (WCAF). Perfect Core Annular Flow occurs when oil and water have the same density. There is a smooth interface between oil and water in Perfect Core Annular Flow. Wavy Core annular flow (WCAF) involves the formation of waves on the surface of the oil core and is thus characterized by a wavy oil-water interface. This is illustrated in Figure 2.5.

Water Oil Wavy oil-water interface

Figure 2.5 Wavy core-annular flow of oil and water (based on Bannwart, 2001 ).

13 It should be noted that PCAF is an idealization and is useful for modeling. WCAF is closer to observations that have been made by researchers. PCAF would thus be expected in the density matched oil-water experiments conducted by Charles et al. ( 1961) even though a wavy interface between oil and water was observed at certain operational conditions. Ripples on the oil core were also observed by Ooms et al. (1984). The 'tiger waves' observed by Joseph et al. (1999) in his bitumen froth experiments are similar to the waves that develop on the oil core in WCAF.

It can be observed from Figure 2.5 that the water layer is thin compared to the rest of the pipe. This is a characteristic of core-annular flow and was observed by Ooms et al. (1984). It can also be observed from Figure 2.5 that the water layer appears to be adjacent to the pipe wall. This was attributed to the formation of waves on the oil core which cause a lifting force on the oil core and was observed by Ooms et al. (1984) in experiments in whicij there was a density difference between the water and the oil.

Different criteria for core-annular flow of heavy oil and water have been proposed by researchers. Oliemans and Ooms (1986) state that core-annular flow will be obtained by 3 crude oils with viscosities greater than 500 mPa· s and densities greater than 900 kg/m • Bannwart (2001) states that crude oils with viscosities greater than 100 mPa· s and densities close to that of water will attain core-annular flow. Joseph (1998) states that core-annular flow of heavy oil will occur with water at the pipe wall if the oil viscosity is larger than 500 mPa· s. Joseph (1998) also states that it is possible to obtain drag reductions of the order of the ratio of the viscosity of oil to that of water in core-annular 5 flow. The ratio of the viscosity of oil to the viscosity of water can be as large as 10 • Bannwart (200 1) states that when the annular fluid is in the turbulent regime and the core fluid is in the laminar regime, the following conditions must be satisfied for core­ annular flow to occur: (2. 3)

14 and e > 0.5 where subscripts 1 and 2 are associated with the core fluid (oil) and the annulus fluid (water) respectively. J.l is viscosity; pis density, J is superficial velocity; D is the pipe diameter and e is the volume fraction of the core.

Bannwart (200 1) showed that this criterion agreed with previous experimental results

(J2 = 1 m/s; J.ltiJ.12 range= 30- 18,000; p1/ p2 range= 0.85- 0.96; D range= 23 to 50 mm). No slip was assumed i.e. both fluids have the same velocity.

A flow map is a representation of sequential flow regimes or patterns and the operational conditions at which these flow regimes occur. It can be observed from the flow map of Charles et al. (1961) that the concentric oil-in-water flow pattern generally occurs between superficial oil velocities of 0.09 m/s and 0.61 m/s at their operational conditions. This suggests that core-annular flow is bounded by upper and lower velocity limits. A critical lower velocity is required to achieve the core-annular flow pattern and beyond the upper velocity limit, the flow regime is lost. Results obtained by Joseph et al. (1999) in self-lubricating flow experiments with bitumen froth indicated the existence of a lower critical velocity limit but no upper bounding velocity was found. Ooms et al. (1984) obtained core-annular flow only above a critical velocity of0.1 m/s. Below this critical velocity, a stratified flow regime was observed with oil in the upper part of the pipe and water below it.

Core-annular flow can be achieved either by the injection or addition of water to oil or by subjecting a water-in-oil emulsion to a shear field (self-lubricating flow).

15 2.4.1.1 Water-Assisted Flow This is the common method of achieving core-annular flow and most of the previous researchers on core-annular flow have employed this method. An inlet device leading into a pipe which consisted of a central tube surrounded by an annular slit was used by Ooms et al. (1984). A nozzle was employed by Charles et al. (1961). Isaacs and Speed (1904) developed an apparatus for achieving core-annular flow. Obstructions (e.g. grooves and ribs) in the pipe were used to impart a helical motion to the combined oil and water flow. However, recent research suggests that the method by which water is injected is not critical to the formation of the lubricating layer (McKibben et al., 2000a). McKibben et al. (2000a) describe various methods including using a Tee junction and injecting the water into the oil through thin longitudinal slots in the pipe wall.

2.4.1.2 Self-Lubricating Flow The objective of self-lubricating flow is to achieve the core-annular flow pattern without the injection or addition of water. This requires water droplets to be dispersed in the oil phase. The water-in-oil emulsion must then be subjected to shear. The dispersed water droplets migrate under the influence of shear to relatively high shear regions in the pipe and coalesce to form an annular water layer around the oil core (Kruka, 1977). This is in agreement with observations made by Neiman (1986) of a thin lubricating water layer around a central plug of bitumen froth in his pipeline flow experiments. The significant difference between water-assisted flow and self-lubricating flow is that the pipe wall is clean in water-assisted flow. Fouling of the pipe wall in self­ lubricating flow of bitumen froth will be explained in detail.

Kruka (1977) describes self-lubricating flow obtained with water-in-heavy oil emulsions (typically 5 - 60% by volume water) in laminar pipeline flow after subjecting them to shear. This is similar to the methods employed in this study in which an emulsion is subjected to a shear rate from a rotating spindle for a certain length of time until the desired flow pattern is obtained.

16 Kruka' s method was demonstrated with three emulsions of water in Midway-Sunset crude oil (emulsions had low shear viscosities of 5000, 44000 and 79000 mPa· s) in a 12.7 mm internal diameter steel tube 1.36 m long. The tube was connected to a pressure vessel suspended from a load cell used to determine flowrate. The vessel was filled with the emulsion and pressurized to the desired level. A discharge valve was used to control the flow. A typical test with one emulsion involved commencing with a low pressure and gradually increasing the pressure until core-annular flow was established. Methods by which the flow patterns were observed are not described. Results indicate a transition from shear thinning laminar flow to an intermittent core flow and finally to a steady core flow. It is not clear what the first two flow patterns represent. The intermittent core flow might be similar to the oil slugs in water flow regime obtained by McKibben et al. (2000a) for heavy oils. Results obtained by Kruka (1977) also support earlier statements about the existence of a critical velocity required to achieve core-annular flow.

Core-annular flow was obtained by Kruka (1977) for shear rates (based on zero-shear 1 viscosity) between 2 and about 5000 s- • It was not stated if the shear rates are the maximum shear rates at the pipe wall. The discovery of shear rate limits for core­ annular flow by Kruka (1977) agrees with the velocity boundaries discovered by Charles et al. (1961). Minimum residence times of 0.1 to about 200 s were also required to achieve core-annular flow. The minimum residence times were found to depend on the viscosity of oil, temperatures, pressure and pipe diameter. It was also stated that longer residence times enable lower shear rates to be used.

Kruka (1977) does not state the temperature at which the tests were conducted. Neiman et al. (1999) suggest that Kruka's method of gradually increasing the pressure cannot be used in the transport of bitumen froth over a distance of about 35 km by pipeline because the pressure drop required to produce the critical shear rates for self-lubricating flow are very high. It appears that their suggestion might be based on economic reasons. Kruka (1977) used emulsions with viscosities comparable to that of bitumen froth even though they were of a lower water content. Bitumen froth has a viscosity of the order of 20 Pa· sat 50°C (Neiman, 1986). Kruka's method might be employed to find the critical

17 velocity required to achieve self-lubricating flow. It should be noted that subsequent research has shown that self-lubricated pipeline transport of bitumen froth is commercially feasible.

Neiman (1986) was the first person to conduct research on self-lubricating flow of bitumen froth. His tests were conducted in a 50 mm diameter pipeloop. Bitumen froth was produced in an experimental extraction unit. Two methods of extraction of bitumen froth were employed to determine possible effects. The Hot Water Extraction (HWE) was conducted at 80°C while Warm Water Extraction (WWE) was conducted at a temperature range from 3 5 to 50°C. It should be noted that the current S yncrude extraction process at Aurora employs lower extraction temperatures of 25 to 35°C (Syncrude, 2003). Batches of bitumen froth (approximately 200 kg) were collected in a hopper and subsequently pipelined. The collected bitumen froth generally had a temperature ranging from 45 to 50°C. This was designated as unheated froth. A steam jacket heat exchanger on the pipeloop was used to heat the bitumen froth to higher temperatures normally for a duration of about five minutes and then maintained at a constant temperature. Froth temperatures used in this study varied from 45 to 95°C. Bitumen froth at a temperature of 80°C was designated as heated froth. Pipelining bulk velocity was varied between 0.2 and 1.0 m/s. Constant velocity tests were conducted at 0.5 mls. It was suggested that this value results in shear rates similar to those obtained in laminar pipe flow on an industrial scale (an industrial pipeline was assumed to have a diameter of 0.6 m). However, it should be noted that the Syncrude operation presently employs an industrial pipeline diameter of about 0.9 m. The water concentration in bitumen froth was varied between 29 and 3 7 wt%. Water concentrations were increased to 62 wt% through injection of water into the pipeline. The studies were usually about 2 hours in duration. Pipeline headlosses were determined by measuring the pressure drop over a fixed length of straight pipe.

Some results obtained by Neiman (1986) are similar to those obtained by Sumner et al. (2003) in which bitumen froth was sheared in a concentric cylinder viscometer. This is clearly illustrated by comparing Figures 2.6 and 2. 7.

18 ~ s 1.0

8'._.I 0.8 Q) '"0·~= e00 0.6 u .....-4·~::s ~ 0.4 ~ = 02

0 2400

Figure 2.6 Measured pipeline hydraulic gradient for bitumen froth as a function of time. Run 019; 35 wt% water, 59 wt% bitumen, 6% solids; T = 80°C; D = 0.0525 m, V

= 0.5 m/s (based on Neiman, 1986).

19 44100

34300 8 3. ';' 24500

0~ E--4 r47oo

360 720 1080 1440

Time (sec)

Figure 2.7 Measured torque of bitumen froth as a function of time in a concentric cylinder viscometer. Run 8-45; T= 45°C, n = 64 rpm: MVIII spindle (Rt = 15.2 mm) (based on Sumner et al., 2003).

It should be noted that the conditions are not the same in Figures 2.6 and 2.7. It can be observed from Figure 2.6 that the initially high pressure gradient shown reduces to a low pressure gradient. It can also be observed from Figure 2. 7 that the initially high torque reduces by orders of magnitude to a low torque associated with the presence of a water layer. This appears to indicate some sort of similarity between self-lubricating flow of bitumen froth in a pipeline and in a concentric cylinder viscometer.

20 A thin water layer around the bitumen core was usually observed by Neiman (1986) during the loop discharge into a head tank corresponding with the low pressure drops associated with self-lubricating flow. This water layer was usually observed to grow in thickness as the water concentration in the bitumen froth increased. The water layer was estimated to be about 2 mm thick (probably visually) at high water concentrations.

Generally, instantaneous low pressure drops were measured by Neiman (1986) at a constant velocity of 0.5 m/s for water contents of about 30 wt% and temperatures ranging from 45 to 57°C. At a higher temperature of 80 °C, an increased water fraction of 3 7 wt% also resulted in similar instantaneous low pressure drops at the same velocity (illustrating the effect of increased water fractions). However, water concentrations between 29 and 35 wt<»lo at a temperature of 80 °C resulted in an initially high pressure drop which reduced to the previously observed low pressure drops after a period of about 55 min. Neiman (1986) concludes that increasing the water content in heated froth (80 °C) from 25 wt% up to about 35 wt% causes a gradual decrease in pressure gradient while there is little variation in pressure gradient in unheated froth (50 °C) at the same conditions. Increasing the water content in both unheated and heated froth above 35 wt% results in low pressure gradients with only a small degree of variation. This behaviour was attributed to the fact that an excess of separable water was available. It was thus concluded that there was little incentive for increasing the froth water content beyond 3 5 to 40 wt% for purposes of transport. As noted above, heated froths (80°C) with water concentrations greater than 35 wt% eventually attained self­ lubricating flow though after a considerable period of time. It appears Neiman (1986) only considered tests which resulted almost immediately in self-lubricating flow (low pressure drops) in his conclusions. Tests which initially had high pressure gradients but eventually resulted in lower gradients were considered as having high pressure gradients. Pressure gradient reductions of up to two orders of magnitude were also obtained with these tests.

21 Joseph et al. (1999) conducted bitumen froth experiments in both a 25 mm pipeloop and a 0.6 m diameter pilot pipeline. In the 25 mm pipeloop tests, water volume fraction varied from 22 to 40% and temperature varied between 35 and 55°C. The 0.6 m diameter pilot pipeline tests involved a temperature range between 45 and 55°C. The duration of these tests were generally longer than those ofNeiman's and varied between 8 and 96 hours in duration. Pressure gradients were measured at different velocities. Low pressure gradients associated with self-lubricating flow were generally observed. 'Tiger waves', which are standing waves on the bitumen core that can be seen when a transparent viewing section is water-wet, were observed during these tests. These reduced pressure gradients were monitored over a considerable period of time and were generally found to be stable. A few tests resulted in large fluctuations and increases in pressure when the velocity was reduced to 0.5 m/s. This caused a failure of the self­ lubricating flow. This failure was attributed to the formation of slugs of bitumen froth which form the continuous core. This is similar to observations made by McKibben et al. (2000a) during the transition from continuous water assisted flow to slug flow. These observations agree with the existence of a lower velocity required to achieve core­ annular flow found by Charles et al. (1961).

In the bitumen froth experiments conducted by Joseph et al. (1999), the pipeline was initially filled with water. Bitumen froth was then introduced at velocities (about 0.3 to 0.9 m/s) greater than critical velocities required to generate self-lubricating flow. Measurements of the critical velocity required to achieve self-lubricating flow were difficult. Critical velocity measurements were made by sequentially reducing the flowrates and thus the pressure drop at a particular temperature. This was done until self-lubricating flow was lost. The velocity at which self-lubricating flow was lost was designated as the critical velocity. It was suggested that these values were close to those required for achieving self-lubricating flow. These critical velocities were observed to decrease with an increase in temperature. Thus, it was concluded by Joseph et al. (1999) that self-lubricating flow was easier to achieve at higher temperatures.

22 Froth samples were taken by Joseph et al. (1999) at different velocities. Free water was separated from the froth and the water content in the froth was determined by distillation. The amount of water in the lubricating layer was calculated to be about 20% of the original water contained in the bitumen froth based on their results. The water layer thickness was estimated to be between 0.3 to 0.4 mm in thickness in the 25 mm diameter pipe based on the froth water composition analysis.

Joseph et al. (1999) correlated the measured pressure gradients with an equation based on the Blasius equation for turbulent flow of water, reasoning that the measured pressure gradients correspond to shear stresses in the water phase only:

(2.4)

where K is a constant of proportionality

This correlation appears to be similar to one developed by Arney et al. (1993) for core­ annular flow. Equation 2.4 predicts that the pressure gradients for self-lubricated flow of bitumen froth increase with a decrease in pipe diameter for a constant velocity.

Joseph et al. (1999) also studied the behaviour of bitumen froth sheared in a 22 mm diameter parallel plate viscometer. The froth contained a water volume fraction of 20 - 22% and was tested at temperatures ranging from 20 to 40°C. Results indicated a general increase in shear stress with shear rate and an approximately constant shear stress was measured at shear rates greater than I 00 s·1 for all temperatures. It was suggested that this constant shear rate obtained was the result of self-lubrication of bitumen froth.

A mechanism based on the properties of the colloidal dispersion of clay particles in froth water was proposed by Joseph et al. (1999). It was suggested that self-lubrication could be achieved in a pipe with an oil layer adhering to the wall (fouling) because the

23 clay particles prevent the bitumen in the core from attaching to bitumen on the wall. However, similar flows have been observed by McKibben et al. (2000b) in water­ assisted heavy oil pipe loop tests where the wall of the pipe was fouled with oil and no fine solids were present.

The fouling of pipe walls by oils is dependent on the material of construction of the pipe. Arney et al. (1996) showed that steel pipes are fouled by oil, and that fouling can be prevented by lining the pipes with cement. Most core-annular flow researchers do not indicate if the water layer formed lies adjacent to the wall or is separated by an oil layer from the pipe wall. K.ruka (1977) states that the annular water layer is formed within approximately 0.6 to 0.9 of the radius from the centreline of the pipe. The method by which this was deduced is not stated. However, this suggests that there might be some heavy oil between the water layer and the pipe wall. This observation also appears to be in agreement with drop migration observations which will be reviewed subsequently. Charles et al. (1961) used a cellulose acetate-butyrate pipe. It was observed that the most viscous oil had a low contact angle with the pipe and thus wetted it. This behaviour was suggested as being responsible for the formation of several flow patterns which had water as the dispersed phase and oil as the continuous phase at low water velocities. Ooms et al. (1984) discovered that the oil fouled the upper part of the pipe when the flow was stopped. Joseph et al. (1999) did not observe a fouling layer but the discrepancy between their pressure drop measurements for water alone in their pipes and the calculated Blasius value was attributed to the presence of a fouling oil layer on the pipe wall. Values closer to the Blasius value were obtained after the pipe was cleaned with clay water. Their results indicate that self-lubricating flow can be achieved even in the presence of some fouling. The low pressure drops obtained with self-lubricating flow were monitored for long periods of time to ensure that there was no growth of the fouling layer. An increase in the fouling layer thickness poses a problem because of the potential for high pressure gradients (Joseph et al., 1997). Schaan et al. (2002) conducted heat transfer measurements during self-lubricating flow of bitumen froth and inferred that the fouling bitumen layer was 2.1 mm thick for flow at a velocity of 1.5 m/s in a 50 mm pipe.

24 It may thus be concluded from the bitumen froth experiments of Joseph et al. (1999) and Schaan et al. (2002) that a fouling oil layer separates the lubricating water layer from the steel pipe wall in self-lubricating flow.

Bitumen froth experiments were conducted in a 25 mm pipeloop by Sumner et al. (2003) for the temperature range 43 to 50°C and a velocity of 1 m/s. The bitumen froth samples contained water at concentrations of 15 to 30 wt%. It was discovered that self­ lubricating flow could not be achieved below a water concentration of 17 wt%. It was also discovered that there was an increase in pressure gradient as the water concentration was decreased. This is in agreement with results obtained by Neiman (1986) and Sanders et al. (2004). It was difficult to maintain constant temperature during these tests. In addition, these tests also required reasonably large volumes of bitumen froth. As a result, other shearing devices were considered for the study to determine if a flow regime resembling the self-lubricating phenomenon would occur. Limited experiments conducted with bitumen froth in a cone and plate viscometer did not yield a substantial reduction in applied shear stress. Shearing of bitumen froth with the MVIII spindle ( 15.20 mm radius) in the Haake RV3 concentric cylinder viscometer was found to result in a significant reduction of the initially high torque. The initial torque value was comparable to the value which would be obtained with bitumen alone. The reduced torque was believed to be associated with the formation of a lubricating water layer within the sheared annular volume. It was thus evident that the self­ lubricating flow behaviour observed with bitumen froth in the concentric cylinder viscometer could be used to model that obtained in the pipe. Experiments with the MVII (18.4 mm radius) and MVI (20.04 mm radius) spindles using the RV3 viscometer did not result in self-lubricating flow (Sumner et al., 2003).

Schramm and Kwak (1988) state that the viscosity of a given bitumen is dependent on the location of the host oil sand and the method of extraction used to obtain this bitumen. Dealy (1988) states that a 'standard' bitumen does not exist and that bitumen

25 must be described in terms of its source, treatment and storage conditions. It would thus be difficult to obtain experimental results with a particular bitumen that would be applicable to all bitumens. The desire for better control of experimental parameters led to modeling bitumen froth as a water-in-oil emulsion (Sumner et al., 2003). Several oils were processed into emulsions and sheared in the concentric cylinder viscometer. Results similar to that previously obtained for bitumen froth were obtained with a Pennzoil N-Brightstock oil. Shellflex 810 oil emulsions did not exhibit behaviour resembling self-lubricating flow in the limited range of measurements in the study.

This study utilizes concepts from the concentric cylinder study of Sumner et al. (2003).

2.5 Particle and Droplet Migration In Viscous Fluids The coalescence of dispersed water droplets under the influence of shear is required for self-lubricating flow. Water droplets which are originally dispersed in a continuous oil phase must thus migrate to high shear regions near the pipe wall (Kruka, 1977; Joseph et al., 1999). A review of observed migration mechanisms of particles and droplets in Poiseuille and Couette flow is thus important. Possible causes of droplet and particle migration will also be examined.

Neutrally buoyant rigid spheres and droplets dispersed in a continuous Newtonian fluid phase have been observed to migrate to equilibrium radial positions in Poiseuille and Couette flow. Neutrally buoyant implies that the dispersed phase has the same density as the continuous phase. This migration is also known as lateral migration because the droplets and particles migrate in the radial direction across streamlines. Segre and Silberberg (1962) conducted experiments to observe the motion of rigid spheres in mixtures of glycerine, 1,3-butanediol and water (viscosity ranging from 17 to 400 mPa· s) undergoing Poiseuille flow at low Reynolds number through a 11.2 mm diameter vertical tube. The density of the spheres and the media were matched. A combination of lamps and prisms were used to provide perpendicular rays of light .to a selected section of the tube. Spheres which coincide with the intersection of the rays of light could then be counted. These counts were used to generate a concentration distribution of spheres.

26 A similar concentration distribution measurement approach was employed by Furuta et al. (1977). The concentration distributions of Segre and Silberberg (1962) showed that a large number of spheres accumulated around one-half of the tube radius. An equilibrium position of about 0.6 of the tube radius from the axis was determined for the rigid spheres. This position was discovered to be independent of their initial position. This is also known as the tubular pinch effect. Similar observations were made by Kamis et al. (1963) in their experiments with single neutrally buoyant rigid spheres, rods and disks. Particles initially placed near the tube wall migrated inwards, while particles near the tube axis moved outwards and finally reached an equilibrium radial position close to one-half of the tube radius. Water droplets in silicone oil (viscosity of 5030 mPa· s) were observed by Goldsmith and Mason (1962) to migrate to an equilibrium position at the tube axis. Similar observations were made by Kamis et al. (1963) for glycerol droplets in a polyglycol mixture. The glycerol droplets were 10 times as viscous as the continuous phase. No migration was observed by Kamis et al. (1963) for liquid droplets (unspecified) 50 times more viscous than the continuous phase. It was suggested that at these conditions, droplet deformation was negligible and the droplets thus exhibited behaviour similar to rigid spheres. Density differences between the dispersed phase and continuous phase in the experiments of Karnis et al. 3 (1963) and Goldsmith and Mason (1962) were of the order of30 kg/m •

Experiments with non-neutrally buoyant spheres (Brenner, 1966) resulted in migration behaviour different from that previously observed with neutrally buoyant spheres. It was discovered that for fluid flowing upwards in a tube, spheres denser than the fluid migrated to the axis of the tube. For fluid flowing downwards in a tube, spheres denser than the fluid migrated to the wall of the tube while spheres less dense than the fluid migrated to the axis of the tube. Similar observations were made by Furuta et al. (1977) who studied migration of rigid spheres in a vertical pipe.

The review by Brenner ( 1966) indicates that migration experiments conducted for Poiseuille flow have a typical droplet radius to pipe radius ratio of about 0.02 to 0.3.

27 No migration of rigid spheres in Newtonian fluids was reported by Gauthier et al. (1971) in their Couette flow experiments conducted in a counter-rotating coaxial cylinder apparatus (radius of inner cylinder: 47.56 mm; radius of outer cylinder: 56.52 mm). In the counter-rotating coaxial cylinder apparatus, the inner and outer cylinders rotate in opposite directions. Karnis and Mason (1967) conducted several experiments involving dispersed phases, such as Ucon oil and water droplets, in silicone oil undergoing Couette flow in a counter-rotating coaxial cylinder apparatus (radius of inner cylinder: 47.5 to 133.5 mm; radius of outer cylinder: 95.6 to 152 mm). It was observed that the liquid droplets migrated away from the cylinder walls attaining equilibrium about halfway between the two cylinders. The direction of migration was found to be independent of the speed and direction of rotation of the cylinder walls. In their Couette flow experiments, Chan and Leal (1981) utilized a coaxial cylinder apparatus in which the inner rotor rotated and the outer cylinder was stationary. It was observed that Newtonian droplets in a Newtonian fluid [e.g. water droplets in Ucon oil (viscosity= 900 mPa·s)] migrated towards an equilibrium position between the centre­ line and the inner rotor. This equilibrium position was found to be dependent on the gap width between the inner rotor and the wall. For the narrow-gap case (corresponding to the large inner rotors), the equilibrium position was near the centre-line. This equilibrium position moved closer to the inner rotor as the gap width increased (i.e. as the radius of the inner rotor decreased). Smart and Leighton (1991) employed a concentric cylinder device (Inner rotor radius of 97.3 mm, outer cylinder radius of 117.4 mm) in which only the inner rotor rotated. It was discovered that the equilbrium position was dependent on droplet size. Droplet diameters ranging from 0.8 to about 3 mm (computed based on given droplet volume) were injected near the outer cylinder using a syringe. Droplet diameters between 0.8 to 20 mm migrated to equilibrium positions between the inner rotor and the centreline of the Couette device. The only exception was a droplet of diameter 3 mm which migrated to a position between the centreline and the outer cylinder. These results appear to be in agreement with the results obtained by Chan and Leal (1981) who used droplets diameters ranging from I to 4 mm. It can be observed that these are reasonably large droplet diameters.

28 Migration in which the particle-particle hydrodynamic interactions become important has not been fully investigated. Other researchers have investigated flows where dispersed phase concentrations are higher and particle-particle interactions become important. For example, Phillips et al. (1992) conducted experiments with concentrated suspensions (volume fraction 0.45 to 0.6) of rigid spheres in a Newtonian oil (viscosity of 4950 mPa· s) in a horizontal concentric cylinder device (inner rotor radius of 6.4 mm; outer cylinder radius of23.8 mm). Spheres with mean diameters of 100 J.tm and 675 J.tm were used. A migration mechanism based on the effect of spatially varying interaction frequency and the effect of spatially varying viscosity was proposed by Phillips et al. (1992).

A shear field causes the irreversible collision of two spheres. A sphere which experiences a greater number of collisions (interaction frequency) on one side than the other will tend to move in the direction of the lower number of collisions. These collisions are proportional to the local shear rate. Particles thus migrate from regions of high shear rate to regions of low shear rate (Phillips et al., 1992). It should be noted that this hypothesis is the opposite of what is generally believed by Joseph et al. (1999) to occur in self-lubricating flow i.e. droplets migrate to regions of high shear near the wall.

Concentration profiles developed by Phillips et al. (1992) from results obtained from the Couette flow experiments revealed a decrease in concentration from the inner spindle to the outer cylinder. It was suggested that this result is in agreement with the expected migration of particles from regions with high shear rate. The rate of migration was also discovered to be proportional to the square of the radius of the particle. No equilibrium position of the spheres is specified as was the case with in the single sphere experiments. Predicted results for Poiseuille flow reveal migration of particles from the wall of the tube to the centre of the tube. This is in contrast with the single particle experiments in which migration of rigid spheres to an equilibrium position about one­ half of the radius was observed. It is not clear if the results obtaip.ed for migration of rigid spheres by Phillips et al. (1992) is directly applicable to the migration of droplets.

29 Most of the single particle migration experiments that have been reviewed were 4 conducted under conditions of small particle Reynolds numbers of the order of 10 • Lateral migration of rigid particles is not predicted for small Reynolds number flows or the Stokes flow regime (negligible or zero inertia). Observations of the migration of rigid particles thus suggests that inertial forces are important in lateral migration (Brenner, 1966; Segre and Silberberg, 1962). Deformation of droplets is important for their lateral migration (Smart and Leighton, 1991 ). A droplet of fluid in a shear field produces a disturbance velocity. This disturbance velocity is symmetric in the absence of deformation. But droplets (low viscosity) deform under shear and cause the disturbance velocity to be asymmetric. This causes droplets to migrate relative to a bounding wall (Smart and Leighton, 1991 ). It may thus be concluded that inertia and deformation are important in the lateral migration of droplets.

Smart and Leighton (1991) also state that the drift (migration) velocity of a droplet normal to a stationary wall decreases inversely as the square of the distance between the droplet and the wall. It can be observed from their results that the largest droplet exhibits the greatest deformation and migrates furthest from the wall. This is in agreement with Ho and Leal (1974) who found that the largest lateral force on a rigid sphere occurred near the wall for Poiseuille flow with a Newtonian fluid continuum.

In smnmary, the rate of migration of droplets in emulsions with low dispersed phase concentrations is most significantly influenced by droplet deformation, inertia effects from the continuous phase and the presence of a boundary.

30 Chapter Three

EXPERIMENTAL EQUIPMENT AND PROCEDURES

The experimental procedure used in the research consists of the following steps:

I. Preparing a water-in-oil emulsion with the aid of a homogeniser. II. Controlled shearing of the produced emulsion in a concentric cylinder viscometer while measuring the resulting torque as a function of time. III. Obtaining photomicrographs of the emulsion for droplet size analysis with the aid of a microscope and a digital camera.

Methods and devices employed by previous researchers to conduct similar experiments will be reviewed in Section 3.1 with the objective of showing that the methods used in this research are consistent with previous work. The procedures used in determining the density and viscosity of the oils will be discussed in Sections 3.2. The experimental procedure described above will be discussed in detail in Sections 3.3, 3.4 and 3.5.

3.1 Introduction Different emulsion preparation devices have been used by investigators. These include a dynamic coalescer (Benayoune et al., 1998), a blender (Nunez et al., 1996), a Sorvall laboratory mixer (Sanchez and Zakin, 1994), and the Glifford-Wood homogeniser model 1-LV (Pal et al., 1992; Yan et al., 1991; Yan and Masliyah, 1995). A newer version of the Glifford-Wood homogeniser model 1-LV, the Greerco Homomixer model1L (Chemineer Inc., North Andover, MA), was used in this research. A review of the experimental procedures used to prepare emulsions reveals that the typical preparation method involves the addition of surfactant to the continuous phase (oil, for a water-in-oil emulsion), accompanied by addition of the dispersed phase and shearing of this mixture with a mixing device (Pal et al., 1992; Yaghi et al., 2001 ). This

31 method is also known as the agent-in-oil method. A different method, in which the dispersed phase was slowly added to a mixture of the surfactant and the continuous phase while shear was applied by a homogeniser, was used by Yan et al. (1991) and Leal-Calderon et al. (1996). A similar method was used in this research.

This research is not concerned with the rheological properties of emulsions. Nevertheless, some of the equipment used for the study of rheological properties of emulsions will be reviewed. These include the Bohlin CS rheometer (Khan, 1996; Pal, 2000); Fann concentric cylinder viscometer (Pal and Rhodes, 1989); Haake concentric cylinder viscometer MIO (Benayoune et al., 1998) and a Contraves Rheomat concentric cylinder viscometer (Yan et al., 1991). The Haake Rotovisco (model RV3, Haake Inc., N.J., USA) and the Haake Rheostress rheometer (model RS 150, Haake Inc, N.J., USA) were used in this research to study self-lubricating flow. Both of these models are concentric cylinder viscometers. The Haake Rheostress rheometer was also used to conduct rheological measurements of the pure oils.

Methods for obtaining photomicrographs and droplet size analyses are not often described in detail in the literature. These methods include an image analyzer (Yan and Masliyah, 1995) and a Zeiss optical microscope equipped with a camera (Pal, 2000). Emulsion samples were obtained by Bhardwaj and Hartland (1994) with a Teflon stick, spread on a glass slide and photographed under a microscope. The droplets were counted with the aid of computer software. The method of Bhardwaj and Hartland (1994) is similar to the droplet size analysis method utilized in this research.

3.2 Characterization of Oils 3.2.1 Density Determination A constant-volume pycnometer was used in this research to determine oil densities. The pycnometer consists of a stopper and a glass bottle. The procedure consists of the following steps: I. The empty dry pycnometer was weighed with the METTLER BB2400 weighing scale.

32 2. The pycnometer was then filled with reverse osmosis water to the top of the stopper until it weeped out of the hole. The pycnometer and the water were then weighed using the METTLER BB2400 weighing scale. This determined the level of fluid to be maintained in the experiment. 3. The measured mass of water was used to calculate the volume of the pycnometer using the density value associated with the water temperature. 4. The HAAKE F3 temperature bath was set to the desired temperature. The pycnometer was then filled with the oil until it weeped out of the hole in the stopper. Once the desired temperature had been reached, the pycnometer was immersed in the temperature bath fluid for about 10 min. 5. The pycnometer was then removed and the outside was wiped dry with Kim­ wipes (Kimberly-Clark Corporation, GA, USA). 6. The pycnometer was weighed and the density of the oil could then be determined.

3.2.2 Viscosity Determination The Haake Rheostress rheometer (model RS 150, Haake Inc, N.J., USA) was used in the determination of the variation of viscosity of the experimental oils with temperature. Calibration of the rheometer with viscosity standards was conducted prior to the viscometry tests and will be discussed in Chapter 4.

The rheometer was always operated at a fixed rotational spindle speed. The experimental procedure consists of the following steps. 1. Compressed air was directed into the air bearing with the aid of a lever. The pressure gauge was maintained at 38 psi. The rheometer, the temperature bath and the computer were then switched on. 2. The temperature bath was set to the desired temperature and the temperature was monitored on the display. 3. The spindle was attached to the rheometer and the cup was firmly attached to the lift by tightening the bolt.

33 4. A single series of experiments (or 'job') was defined using the RheoWin Pro Job Manager version 2.93 (Haake Inc., NJ, USA) and comprises the following tasks: a) Calibrating the lift distance for the spindle by locating and setting the zero point (the point at which the spindle touches the base of the cup), separating the spindle and the cup after the zero point has been located, and displaying a message indicating that the cup is ready to be filled. The cup was then filled to the designated level based on the spindle type. b) Monitoring the torque as a function of time for a period of about 45 min to 1 hour at a low rotational spindle speed to ensure that the emulsion has reached thermal equilibrium. The low rotational speed (2 rpm) was selected to minimize viscous heating of the oil. Viscous heating effects will be explained in Chapter 4. c) Monitoring the torque as a function of time for a period of about 1 hour at the desired spindle speed. d) Increasing the spindle rotational speed in a stepwise manner and recording the corresponding torque. A period of about 10 min was allowed at each spindle speed. e) Decreasing the spindle rotational speed in a stepwise manner and recording the corresponding torque. A period of about 10 min was allowed at each spindle speed. This was done to check for any time-varying behaviour. This completed the rheological measurements. f) The generated torque-time data were then saved in a computer file for interpretation. 5. The shut-down procedure is the opposite of the start-up procedure described in step 1 with the compressed air turned off last.

3.3 Emulsion Preparation The experimental set-up used for the preparation of emulsions consists of a Greerco homomixer model 1L (Chemineer Inc., North Andover, MA) and a variable autotransformer (Variac) (Staco Energy Products Co. Dayton, OH., USA) which regulates the power supply. The Greerco homomixer is illustrated in Figure 3 .1.

34 Turbine

Upper Plate

Mixing head

Lower plate

Figure 3.1 Schematic diagram of the Greerco homomixer model 1L. A beaker containing the viscous oil is placed so that the lower plate and mixing head are fully submerged.

A considerable amount of work was conducted to develop a procedure that would produce a reproducible emulsion with a narrowly sized distribution of fine droplets. This involved varying the positions of the upper plate, lower plate and mixing head. Beaker volumes and emulsion mixing times were also varied. Stability tests were also conducted on the produced emulsions, which will be discussed in detail in Chapter 4. The optimum procedure developed involved setting the lower plate of the homomixer at a fixed distance of 1 nun from the base of the beaker with the aid of the supporting bolt. The lower plate was also set at a fixed distance of about 5 nun from the mixing head. The upper (deflector) plate was positioned above the fluid level so that it did not affect the emulsion preparation.

35 The steps followed in preparing the water-in-oil emulsions are described below. The emulsions were prepared at room temperature for most of the experiments. Selected tests were performed where the oil and water were preheated separately before mixing to determine if the character of the emulsion was affected. The results from these tests were found to agree with results obtained using the previous method without preheating. 1. The required masses of oil and reverse osmosis water were set to provide the 3 target water concentration and a total mixture volume of 364 cm • The oil and water were weighed into 600 mL and 400 mL PYREX beakers, respectively using a METTLER BB2400 weighing scale. 2. The mixing head of the homogeniser was immersed in the beaker of oil. 3. The variable autotransformer was set to 120 V. The Greerco homomixer was switched on and variable autotransformer was set to 70% of maximum output voltage. 4. The beaker of oil was then rotated continuously while reverse osmosis water was slowly added. After the water had been transferred, shearing was continued for 1.5 to 2.5 min to ensure adequate mixing was achieved. 5. Approximately 100 ml of the uppermost part of the emulsion was then disposed off to ensure that a homogeneous emulsion sample was available for subsequent measurements. A sample of the remaining emulsion was then taken for droplet size distribution analysis with the aid of a pipette dipped into the centre of the beaker. The emulsion in the beaker was then used for the shearing experiments.

3.4 Droplet Size Analysis A NIKON COOLPIX 990 digital camera (8-24 mm) mounted on a Nikon Metallurgical Microscope Eclipse ME 600L (Nikon Canada Inc. Instrument Division, Winnipeg, Man.) was used to obtain photomicrographs of the samples obtained from the water-in­ oil emulsions. The procedure for obtaining microphotographs and conducting droplet size analysis will be described subsequently.

36 A Pasteur pipette (VWR Scientific Inc., West Chester, PA) was dipped into the emulsion and a small sample was withdrawn and placed onto a 76.2 x 2S.4 mm microscope slide (VWR Scientific Inc., West Chester, PA). An 18 x 18 mm square cover glass (VWR Scientific Inc., West Chester, PA) was then placed over the sample. The sample was normally allowed to sit undisturbed for a period of 1S to 20 min. Motion of the droplets was observed when the duration was insufficient. The slide was then placed under the objective lens of the microscope. The microscope was always switched to the DIA mode (diascopic). This mode allows transmission of light through the sample.

The 1OX eyepiece lens was used. The sample was initially observed using the 1OX objective lens. A representative region was selected and then the sample was viewed using the SOX objective lens. This was done to ensure that the droplets were large enough to be analysed in subsequent steps of the procedure. The knob controlling light passing through the sample was adjusted to provide maximum lighting. Then with the same SOX magnification objective lens, the sample was viewed with the camera using full zoom (corresponding to F6.3 on the camera). A photograph of the sample was then obtained. Using the same settings as above, a photograph was taken of the Nikon objective micrometer to permit calibration of the image size analysis software.

Droplet size analysis was conducted with the aid of the Simple PCI software (Version 4.0.0, Compix Inc. Imaging Systems, PA, USA). The procedure involved loading the photomicrograph into the Simple PCI program and calibrating it with the digital photomicrograph of the micrometer. Lines representing diameter were manually drawn across each droplet. Droplet diameter data were thus generated using the software, and then exported into Microsoft Excel for further graphical interpretation.

3.5 Couette Flow Test Equipment The basic principles of the concentric cylinder viscometer used for shearing the water­ in-oil emulsion have been explained in Chapter 2. The Haake Rotovisco viscometer was used for majority of the emulsion shearing tests conducted in this study. The Haake

37 Rheostress rheometer was used for experiments requiring high torque measurements because it has a maximum torque three times that obtainable with the Haake Rotovisco viscometer. The procedures for using the two viscometers are slightly different. It is thus necessary to review both procedures. Reproducibility of the sheared emulsion results obtained with the Haake RV3 viscometer was established with a few tests before the Haake RS 150 rheometer was used. The specifications of the Haake Rotovisco viscometer and the Haake Rheostress rheometer are shown in Tables 3.1 and 3.2 respectively.

3.5.1 Haake Rotovisco Viscometer

The experimental apparatus comprises the following:

1. The Haake F3 temperature bath (Haake Inc., N.J., USA) which is used to control temperature. ii. The Haake Rotovisco viscometer (model RV3, Haake Inc. NJ. USA) with a model MK500 measuring head unit. It has a cup and three available cylindrical spindles (models MVI, MVII and MVIII). A thermometer is located in the jacket enclosing the cup to indicate the temperature at the viscometer. The dimensions of the cup and spindle are shown in Table 3.1 iii. The X-Y chart recorder records the torque exerted on the rotor (spindle) as a function of time.

Table 3.1 Haake RV3 Viscometer Specifications.

Spindle MVIII MVII MVI Radius, Rt (mm) 15.2 18.4 20.04 Length {mm) 60 60 60 Cup Radius, R2 (mm) 21 Head MK500 Maximum torque (Nm) 0.049

38 The steps followed in shearing the water-in-oil emulsions with this viscometer are stated below: 1. The Haake F3 temperature bath was set to the desired temperature prior to preparing the emulsions with the homogeniser. The thermometer was monitored to ensure that the desired temperature had been reached. 2. The cup was then filled to the designated level (based on the spindle type) with the previously prepared emulsion. The emulsion was initially sheared at a low speed to promote thermal equilibrium with minimal viscous heating. The torque was monitored as a function of time on the chart recorder. Thermal equilibrium is assumed to have occurred when the torque remains constant with respect to time at a fixed shear rate. This normally required a period of about 30 min. The concept of viscous heating will be explained in detail in Chapter 4. 3. The first experiment conducted at a given temperature normally involved increasing the spindle rotational speed in stepwise increments until self­ lubricating flow was obtained (Stepwise Spindle Speed procedure). It should be noted that discrete spindle speeds were used in this research for consistency. These spindle speeds are 8, 16, 22.6, 32, 45.2, 64, 90.5, 128, 181, 256, 362, 512 and 724 rpm. Each speed was normally maintained constant for a period of about 5 to 10 min before adjusting to the next higher speed. Subsequent experiments were then performed at specific constant spindle speeds (Constant Spindle Speed procedure) to verify the lowest spindle speed which resulted in self-lubricating flow. The torque was monitored on the chart recorder as a function of time normally for a shearing duration of about 30 to 60 min. 4. After shearing, a sample of the emulsion was taken from the cup for subsequent droplet size analysis.

3.5.2 Haake Rheostress Rheometer The experimental apparatus consists of the following: i. The Haake Rheostress rheometer (model RS 150, Haake Inc., NJ.,USA). It has three available cylindrical spindles (models Z31 Ti, Z38Ti and Z41 Ti). The cup and spindle dimensions are shown in Table 3 .2.

39 u. The Haake DC30 temperature controller. iii. The Thermo Haake K1 0 temperature bath.

IV. A computer for data acquisition.

Table 3.2 Haake RS 150 Rheometer Specifications.

Spindle Z31Ti Z38 Ti Z41Ti

Radius, Rt (mm) 15.725 19.01 20.71

Length (mm) 55 55 55

Cup Z43

Radius, R2 (mm) 21.667

Maximum torque (Nm) 0.15

The rheometer was always operated in the Controlled Rate mode. The experimental procedure consists of the following steps: 1. Compressed air was directed into the air bearing with the aid of a lever. The pressure gauge was maintained at 38 psi. The rheometer, the temperature bath and the computer were then switched on. 2. The temperature bath was set to the desired temperature and the temperature was monitored on the display. 3. The spindle was attached to the rheometer and the cup was firmly attached to the lift by tightening the bolt. 4. A programmed set of operating instructions was defined using the Rheo Win Pro Job Manager version 2.93 (Haake Inc., NJ, USA). The following tasks were performed. a) Calibrating the lift distance for the spindle by locating and setting the zero point (point at which the spindle touches the

base of the cup), separating the spindle ~d the cup after the zero point has been located, and displaying a message indicating that cup is to be filled.

40 b) Monitoring the torque as a function of time for a period of about 45 min to 1 hour at a low spindle rotational speed. This is to ensure that the emulsion had reached thermal equilibrium with minimal viscous heating. c) Monitoring the torque as a function of time for a period of about 1 hour at the desired spindle speed (Constant Spindle Speed procedure). A few tests were also conducted using the Step-wise Spindle Speed procedure described in Section 3.4.1. The discrete spindle speeds described in 3 .4.1 were also used. d) Saving generated torque versus time data as a computer file. 5. After the run, the shut-down procedure is the opposite of the start-up procedure described in step 1 with the compressed air turned off last.

It should be noted that Step 4, which is the programming step, is the only difference between the procedure used in shearing the emulsion and that for conducting rheological measurements. All other steps are identical.

41 Chapter Four

RESULTS AND DISCUSSION

4.1 Continuous Phase (Oil) Characterization Two model oils and coker-feed bitumen (vacuum tower bottoms, Syncrude Canada Ltd.) were used in this study. The model oils were N-Brightstock (Calumet Lubricants, Indianapolis, IN) and Shell flex 810 (Shell Canada Ltd., Calgary, AB). The procedure by which the density and viscosity of the oils were determined has been described in Chapter 3. The results from the density and viscosity measurements will be analysed and compared to available literature in this section.

4.1.1 Density measurements Density measurements of the oils were conducted using a constant volume pycnometer as previously described in Chapter 3. No density measurements were conducted for the bitumen.

Experimental results from the density measurements of N-Brightstock oil for the temperature range 10 to 50°C are provided in Table A.1 in Appendix A. The results are also plotted in Figure 4.1. It can be observed from Figure 4.1 that oil density decreases with an increase in temperature. This is in agreement with the general expected behaviour of liquids.

Three measurements were conducted at 30°C to ascertain the degree of accuracy and precision of the experimental procedure. An average density of approximately 919.5 kglm3 and a standard deviation of 0. 79 kglm3 was determined for these measurements.

42 The experimental data were fitted with the aid of Microsoft Excel to the following polynomial equation as shown in Equation 4.1.

p = -Q.0095T2 + 0.228IT + 920.87 (4.1)

where p is density in kg/m3 and T is temperature in °C.

924

,.. • 03nsity rreast.rerra1s 1--- ~ - ~}OJT'ial fit

.-..918 ~ C") ...... E ._..~916 ~ ~ ·;; 914 \ c Cl) c 912 \

910 \ \

I I I I 0 10

Figure 4.1 Measured N-Brightstock oil density as a function of temperature . (temperature range of 10 to 50°C).

43 A correlation coefficient of 0.996 was calculated for the data fit indicating that the curve fits the experimental data quite well.

A nominal density of 930 kglm3 at 60<>p (15.56°C) was given by the N-Brightstock oil supplier. A density of 922.11 kglm3 at 15.56°C was determined from Equation 4.1. This value is 0.85% less than the supplier's value.

Experimental density measurements obtained for Shellflex 810 oil for temperatures from 10 to 50°C are provided in Table A.2 in Appendix A and plotted in Figure 4.2. It can be observed from Figure 4.2 that the Shellflex 810 oil exhibits a decrease in density with increasing temperature. This is in agreement with the observation previously described for N-Brightstock oil. It can also be observed that for any given temperature, the Shellflex 810 oil has a lower density than the N-Brightstock oil.

The experimental data shown in Figure 4.2 were fitted to the following linear equation with the aid of Microsoft Excel.

p = -0.5676T +903.77 (4.2) where p is density in kglm3 and T is temperature in °C.

A correlation coefficient of 0.9996 was determined for the linear fit. This linear Equation 4.2 differs from the polynomial Equation 4.1 previously obtained for N­ Brightstock oil. A linear relationship similar to Equation 4.2 between density and temperature was also obtained for crude oil by Benayoune et al. (1998).

A nominal density of 894.7 kglm3 at 15°C was given by the Shellflex 810 oil supplier. A density of 895.26 kglm3 at 15°C was computed with Equation 4.2. This value is 0.06% greater than the supplier's value.

44 ~~------~

895 +------~------I • Density measurements -Unearfit

~800+------~~------~ (") .._E C) ._..~ ~~+------~------~ ·;;.. c: CD c~+------~~------~

875+------~-----~

870T------.------~------~------~------r------~ 0 10 20 30 40 50 60 Temperature (°C)

Figure 4.2 Measured Shellflex 810 oil density as a function of temperature (temperature range of 10 to 50°C).

45 4.1.2 Viscosity Measurements A Cannon S600 viscosity standard (Cannon Instrument Company, USA) was used to verify the accuracy of the Haake RS 150 rheometer prior to viscosity measurements (viscometry) of the oils. Viscosity measurements of the oils were conducted with the aid of the Haake RS150 rheometer. Torque (exerted by the oil on the rotating spindle) and spindle rotational speed data were obtained from these measurements. These data were fitted using the Newtonian rheological model equation for a fluid in a concentric cylinder geometry (Equation 2.1 ). The data were also fitted using the Bingham rheological model (Equation E.15; Appendix E), to indicate the presence of yield stresses in the oil.

A temperature gradient will exist for some period of time between the oil and the temperature bath if they are not at identical temperatures. The resulting temperature profile within the viscous oil may contribute to errors in viscometry measurements. The presence of this temperature gradient as well as viscous heating were found to affect the quality of the data. Figure 4.3 shows the effect of these two factors on the viscosity measurements obtained. Figure 4.4 shows viscosity measurements in which steps were taken to reduce the effect of the two factors. Figure 4.4 indicates an elimination of the time varying effect associated with viscous heating and presence of a temperature gradient.

46 80 .. 70

.-.60 -!! "C !so ·u~ .240 CD >.... • l\t1easured torque data .!!30::s CS) - Unear Newtonian roodel tit c c( 20

10

0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Torque/Length (N-mlm}

Figure 4.3 Newtonian model fit for measured torque and angular velocity data ofN­

Brightstock oil obtained using Z38 Ti spindle (R1 = 19.01 mm) at a temperature of 50°C (Correlation coefficient of linear Newtonian model fit, R2 = 0.99). Spindle speed range = 1 to 700 rpm.

47 1.2.------~

1.0

.-.. ..!! ~0.8 ...... ·c:;~ .20.6 ~... .! :::s ct»0.4

0.2 + Measured torque data - Unear NeNtonian rrodel fit

0.0+-----~------~----~----~------~----~------~----~ 0.0000 0.0010 0.0020 0.0030 0.0040 0.0050 0.0060 0.0070 0.0080 Torque/length (N-m'm)

Figure 4.4 Newtonian model fit for measured torque and angular velocity data ofN­

Brightstock oil obtained using Z31 Ti spindle (R1 = 15.725 mm) at a temperature of 50°C (Correlation coefficient of linear Newtonian model fit, R2 = 1). Spindle speed range = 2 to 10 rpm.

48 4.1.2.1 Effect of Temperature Gradient The viscosity measurements (viscometry) were conducted at a constant preset temperature. It was thus necessary to ensure that the entire oil sample contained in the cup of the viscometer is at this temperature. The torque (exerted by the oil on the spindle) is a function of the viscosity of the fluid. This is valid at conditions of constant temperature, spindle speed and fixed spindle and cup dimensions. This relationship between the torque and the viscosity is also evident from Equation 2.1.

The oil sample was allowed to achieve thermal equilibrium with the temperature bath by rotating the spindle at a low speed for a certain duration. The low speed was used to minimise viscous heating which will be discussed in Section 4.1.2.2. The duration required to achieve thermal equilibrium was determined as approximately 1 hour from tests with N-Brightstock oil. Measurement of the torque as a function of time was used as an indication of when the temperature of the oil was constant. Figure 4.5 shows an example of a plot of the measured torque as a function of time during this temperature control test. It can be observed that the torque initially decreases with time. There is a variation in the rate of the temperature change of the oil during this period which is associated with the slope. The torque measurements become constant after a duration of about 1 hour, which indicates that the oil sample has reached thermal equilibrium with the temperature bath.

49 6000.------~

• 5000 •

4000 • -E • I • z • ~ • ;- 3000 •• :::J •• e- •••• ~ .,.~11111 2000 ~ ~ 111111111111111111111111111111111111111111111111111111111111111111111 14------lr~

rapid change in slope 1000

o~---.-----r----~---~r---~~---~----~----~ 0 500 1000 1500 2000 2500 3000 3500 4000 Time (sec)

Figure 4.5 Measured torque data ofN-Brightstock oil as a function of time using Z31

Ti spindle (R1 = 15.725 mm) at a constant spindle speed of 50 rpm and a temperature of 50°C.

50 4.1.2.2 Effect of Viscous Heating Shearing of a viscous fluid in the viscometer causes the layers of the fluid to move against each other. This results in the production of heat. This heating effect is known as viscous heating or viscous dissipation. Significant viscous heating may result in an increase in fluid temperature and an associated decrease in fluid viscosity. This could result in measurement errors.

An approximate model was developed for viscous heating in oil. The oil was assumed to be a perfect insulating medium (no heat loss to the environment). It was also assumed that the work done by the fluid in exerting a torque on the rotating spindle is completely converted into heat. This heat raises the fluid temperature by ~T °C. The following equation was derived:

(4.3)

where ~ T is the change in fluid temperature, J.1 is the fluid viscosity, L is the length of the spindle, ro is the rotational angular velocity of the spindle, R1 is the radius of the spindle, R2 is the radius of the cup, m is the mass of the fluid, Cp is the heat capacity of the fluid and t is the duration of shear. Details of the derivation of Equation 4.3 are provided in Appendix C.

According to Equation 4.3, the temperature difference in the fluid resulting from viscous heating is directly proportional to the square of the angular velocity of the spindle. This is in agreement with the equations found in the literature. Joseph et al. (1999) report that the temperature increase due to frictional heating in their pipeline experiments was proportional to the square of the velocity. Hashimoto et al. (2001) discovered a relationship between this temperature increase and the square of the shear rate for a concentric cylinder viscometer in which only the outer cylinder rotates.

51 Equation 4.3 predicts that the viscous heating effect increases as the spindle radius is increased for the same spindle speed. This is illustrated in Table 4.1 for the Z31 and

Z41 spindles. The viscous heating effect (represented as ~T) is shown for spindle speeds varying from 2 to 10 rpm and temperatures varying from 10 to 50°C for each spindle. The duration of shear is fixed at I hour.

N-Brightstock oil was used as a sample oil for calculations using Equation 4.3. The heat capacity for N-Brightstock oil was not available. A heat capacity value of2.09 kJ/kg °C based on bitumen (Litzenberger, 2002) was therefore used as a reasonable estimate of the heat capacity ofN-Brightstock in the calculations. It can be observed from Table 4.1 that at similar spindle spee~s and temperature, the larger diameter Z41 spindle exhibits a higher viscous heating effect than the Z31 spindle. The Z31 spindle was thus used in the viscometry of the oils because of the reduction in viscous heating effects.

Table 4.1 Effect of spindle size on viscous heating of N-Brightstock oil. Viscosity at

10°C = 64.97 Pa· s, Viscosity at 50°C = 1.04 Pa· s; R2 = 21.667 mm; Z31 spindle (Rt =

15.725 mm), Z41 spindle (R1 = 20.710 mm), L = 60 mm, t = 1 hour.

Spindle Temperature Spindle speed (rpm) \C) Predicted AT rc) Z31 0.037 Z41 2 10 1.3 Z31 0.92 Z41 10 10 33 Z31 0.00060 Z41 2 50 0.021 Z31 0.015 Z41 10 50 0.53

52 4.1.2.3 Verification of Viscometer Accuracy A Cannon S600 standard was used to verify the accuracy of the Haake RS 150 rheometer prior to viscometry of the oils. Viscometer standards are generally considered to be Newtonian fluids. The Z31 spindle was used in this verification. These verification measurements were conducted at 20°C. Linear data fits obtained using the Newtonian and Bingham model equations are provided in Figures A.1 and A.2 in Appendix A.

The Cannon S600 standard has a given viscosity value of 2.165 Pa.s at 20°C. A Newtonian viscosity of approximately 2.310 Pa.s was computed from these measurements. This is greater than the given value by 6. 7%. This large error might be due to the age of the standard. An intercept yield stress of only 0.35 Pa was obtained with the Bingham model fit (Equation E.13, Appendix E) indicating that the instrument exhibited only a small offset.

4.1.2.3 Rheology of Oils The Z31 spindle was used in the viscometry of the oils because of its small viscous heating effect. The use of the Z31 spindle also enabled comparison with majority of the results from the self-lubricating flow experiments. This will be illustrated further in Section 4.8. Viscosity measurements were conducted using a spindle speed range of 2 1 to 10 rpm (corresponding to a maximum shear rate range of 0.89 to 4.43 s- ). This spindle speed range was selected to minimize the effect of viscous heating after several experiments with higher spindle speed ranges were conducted.

Viscometry results obtained for N-Brightstock oil are provided in Table A.3 in Appendix A. A decrease in oil viscosity with increasing temperature is indicated in Table A.3. The computed yield stresses obtained at low temperatures of 10 and 15°C appear to be greater than that obtained from calibration with the Cannon S600 standard (0.35 Pa). These yield stresses are 6.98 and 1.97 Pa respectively. The computed yield stresses for the temperature range 22.5 to 50°C are close to the value of 0.35 Pa obtained for the Newtonian Cannon S600 standard. N-Brightstock oil will thus be

53 considered as a Newtonian fluid for the temperature range of 22.5°C to 50°C. The Newtonian viscosities in this temperature range are plotted in Figure 4.6. An example of a Newtonian model fit to the measured torque and spindle speed data has been shown in Figure 4.4.

16

14 •\ • Viscosity rmasurerrents - -Equation 4. 7 (Andrade equation) 12 \

<1!10 \ -cu a. -~ ~ 8 \ fA 0 (.) fA >·- 6 \

4 ~

2 ~ ~

0 I 20 30 40 50 60 Tef11JEHCdure \C)

Figure 4.6 Viscosity measurements of N-Brightstock oil as a function of temperature (temperature range of 22.5 to 50°C). Line represents oil viscosities calculated with Equation 4.7.

54 Schramm and Kwak (1988) used the Andrade equation to represent viscosity data as a function of temperature:

(4.4) where J.1 is the viscosity (Pa· s) and T is the absolute temperature (K). A and b are constants.

Equation 4.4 can be rewritten as

In p = Yr +In A (4.5) The experimental data for· N-Brightstock oil shown in Figure 4.6 can thus be represented using Equation 4.5. The constants A and b can thus be determined. This plot is shown in Figure A.3 in Appendix A. The least squares fit shown in Figure A.3 has a correlation coefficient of 0.999.

The least squares fit for the data shown in Figure A.3 is given by the equation

In p = 9211.Yr- 28.507 (4.6)

A is thus determined as 4.16 x 1o- 13 Pa· s and b as 9.2112 x 103 K.

The experimental data can thus be represented by the following equation

9211.2 Jl = 4.16 X }0-l3 e-T- (4.7)

Equation 4. 7 is plotted as a curve in Figure 4.6. It can be observed that Equation 4. 7 adequately describes the experimental data.

55 Viscometry measurements for N-Brightstock oil samples were also conducted with the Z38 and Z41 spindles at 30°C. Computed Newtonian viscosities of 5. 78 and 5.02 Pa· s were obtained. The computed Newtonian viscosities were found to be lower than that previously obtained with the Z31 spindle (6.54 Pa· s). Hysteresis was also observed in the data. This is probably due to more pronounced viscous heating effects associated with the use of larger spindles as previously illustrated in Table 4.1 It would therefore · be incorrect to accept these computed viscosity values. This further validates the use of the smaller Z31 spindle in obtaining the viscosity measurements.

A kinematic viscosity of 1824 eSt at 40°C (0.0001824 m2/s) for N-Brightstock oil was provided by the manufacturer. The density of N-Brightstock oil at 40°C was not 3 provided. Thus, using the density (915.75 kg!m ) and viscosity values (2.42 Pa· s) at 40°C obtained in this research, a kinematic viscosity of 0.00265 m2/s was computed. This value (0.00265 m2/s) is about 45% higher than the given value (0.001824 m2/s). In spite of the high percentage difference, there is confidence in the procedure and the results obtained in this research.

The viscometry results obtained for Shellflex 810 oil are presented in Table A.4 in Appendix A and plotted in Figure 4.7. Oil viscosity is observed to decrease with an increase in temperature. This is in agreement with the observation for N-Brightstock oil. An example of a Newtonian fit for Shellflex 810 oil is provided in Figure A.5 in Appendix A.

56 16

14 • \t1scmity mea5l.I'EI'Tiel r-- H - Equatim4.8 (Praade EQ.Btim)

12 \

-. Cl!10 \ ca .._..a. \ \

4 \

2 ~ ~ ... ~ 0 T I 0 10 20 30 40 50 00 Terrperature \C)

Figure 4.7 Viscosity measurements of Shell flex 810 oil as a function of temperature (temperature range of 0 to 50°C). Line represents oil viscosities calculated with Equation 4.8.

57 A linear plot of In J! against 1/T is shown in Figure A.4 in Appendix A reveals that the data for Shellflex 810 can be similarly described by the following equation.

6896.6 ,U = 1.354 X 1o-to e-T- (4.8)

Equation 4.8 is plotted as a curve in Figure 4. 7.

A viscosity value of 0.458 Pa· s at 40°C is provided by the manufacturer for Shellflex 810 oil. This value is about 10% higher than the corresponding viscosity value of 0.502 Pa· s obtained in this research.

Viscometry of bitumen was conducted only at 50°C. The Newtonian fit to the experimental data is shown in Figure A.6 in Appendix A. A Newtonian viscosity value of 25.63 Pa· s was computed from the data. A yield stress of about 0.37 Pa was computed using the Bingham model. Comparing this low yield stress value with those previously obtained with the viscosity standard fluid indicates that the bitumen is Newtonian. The computed Newtonian viscosity of 25.63 Pa· s is greater than the value of 4.69 Pa· s determined by Schramm and Kwak (1988) for bitumen at the same temperature of 50°C. The higher viscosity computed for the bitumen used in this research is believed to result from the fact that a different feedstock was used for this study (vacuum tower bottoms as opposed to atmospheric distillation bottoms).

4.2 Emulsion Preparation and Characterization An initial objective in this study was to develop an adequate emulsion preparation technique, which should produce a stable emulsion characterized as having small droplets with a narrow droplet size distribution. Different emulsion preparation techniques were tested using the Greerco 1L Homomixer. This involved varying beaker sizes, the distance of the lower plate from the base of the beaker and the distance of the upper plate above the emulsion as previously described in Chapter 3. The optimum procedure has also been described in Chapter 3. This procedure was found to produce a homogeneous emulsion based on visual inspection. The stability of the emulsion was assessed as well as the variability between two emulsions prepared separately.

58 4.2.1 Emulsion Stability The stability test was conducted using a 25 wt% water-in-N-Brightstock oil emulsion. This was chosen as a typical emulsion based on previous work by Sumner et al. (2003). Samples were drawn from the emulsion (as described in Chapter 3) immediately after preparing the emulsion and again four hours after preparation. A duration of four hours was chosen as slightly greater than the maximum required time for a complete Couette flow shearing experiment with the viscometer. The samples which were drawn immediately after preparation were designated as Samples A, B, C and D. Those drawn after four hours were designated as E, F, G and H.

Typical droplet size distributions for samples taken immediately (Sample B) and after four hours (Sample F) are shown in Figure 4.8. Logarithms of the droplet diameters were computed and plotted as the abscissa to include the large droplet diameters at the tail of the distribution.

Results from the analysis of these tests are summarized in Table 4.2. The number, length mean droplet diameter or arithmetic mean droplet diameter is defined by the following equation.

(4.9) where ni is the frequency of a droplet of diameter di and N is the total number of droplets.

The Sauter mean diameter has been used to characterize emulsions by Pal (2000) and Otsubo and Prud' homme (1994) and is shown in Table 4.2. It is defined by the equation

""n.d~ d _L.J,, (4.10) s,v - "" d 2 L.J n; ;

59 100 ..-. e.,~ go CD N ·~ 80 ,CD § 70 >- CJ ceoCD ::J C" ! 50 ..... -+-Sample B (immediate) CD .2!!.., 40 ---Sample F (after 4 hrs) .! e3o ::J CJ ~ 20 ;:: ca "i 10 0:: 0 -1 -0.5 0 0.5 1 1.5 Logarithm of droplet diameter (IJm)

Figure 4.8 Comparison of the droplet size distributions of 25 wt% water-in-N­ Brightstock oil emulsion Samples B and F. Sample B was taken immediately after preparation while Sample F was taken after allowing the emulsion to stand for 4 hours.

60 Table 4.2 Mean droplet diameter and standard deviation data for 25 wt% water-in-N-Brightstock oil emulsion stability tests.

Sample Group Mean standard Group average standard droplet Sauter mean deviation droplet deviation Number of droplets diameter diameter ds, v based on dN,L diameter based based on dN,L Sample counted dN,L (pm) (pm) (pm) on dN.L (pm) (pm)

A 243 2.0 3.1 1.0

B 261 1.6 4.3 1.3

\ 0\ c 296 1.4 2.9 0.9 ~ D 319 1.6 13.3 2.0

1.7 0.3 I

' E 363 1.7 3.4 1.0

F 475 1.9 10.5 2.2 G 337 1.7 12.4 2.0 H 279 1.5 4.2 1.2

1.7 0.2 From Table 4.2, the group average droplet diameters based on dN,L of samples A to D and E to H appear to be similar. It can thus be concluded that the emulsion remained reasonably stable within the duration of the test.

The group standard deviations appear to be large compared with the mean droplet diameters. This is due to the fact that majority of the droplets fall into a narrow size distribution but there is a small fraction of large droplets.

The Sauter mean diameters appear to be very different from the arithmetic mean droplet diameters for Samples D, F and G. The large droplets in the droplet size distribution, though few, are expected to significantly influence the Sauter mean diameter since the diameters are cubed according to Equation 4.1 0.

A 30 wt% water-in-N-Brightstock emulsion was also prepared and allowed to stand for about 24 hours. Droplet size distributions for samples taken immediately after preparation and after 24 hours are shown in Figure 4.9. The droplet size distributions appear to be very similar despite the 24 hours of aging. An arithmetic mean droplet diameter and standard deviation of 2.6 J.lm and 2.4 J.lm respectively were computed for the droplet size distribution of the emulsion sampled immediately. A mean droplet diameter and standard deviation of2.4 J.lm and 1.7 J.lm, respectively, were computed for the droplet size distribution of the sample taken after 24 hours. There is an 8% difference between these two mean droplet diameters.

The similarity of the results from Figure 4.9 further confirm the stability of the emulsion within the maximum duration of experiments performed in this study.

62 100 .-... .._.~90 CD N ·~ 80 CD "'C 70 ::::sc >. u -+- O'oplet diameter data (t =0) coo CD ::::s +-O'oplet diameter data (after 24 hrs) C" ~50 CD ;:;40> .!! ::::s E30 ::::s u ~20 ;:; "i"' 10 0::

0 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 logarithm of droplet diameter (IJI11)

Figure 4.9 Comparison of droplet size distributions of30 wt% water-in-N-Brightstock oil emulsion sampled immediately (t = 0) and after 24 hrs.

63 4.2.2 Assessment of Droplet Size Distribution Reproducibility and Variability The previous section 4.2.1 involved assessing the stability of an emulsion after a significant duration of aging and it was concluded that the emulsion was reasonably stable. It was also necessary to assess the reproducibility when two different emulsions with the same constituents were prepared and analysed using the same methods. The level of variability of the emulsion droplet size distribution within the preparation beaker was also tested.

A 30 wt% water-in-N-Brightstock oil emulsion was chosen to represent a typical emulsion. Two separate emulsions were prepared. For a particular emulsion, two samples were drawn from the top of the emulsion after a small portion had been poured off. Most of the emulsion was then poured out and two more samples were drawn from the lower portion of the beaker. The initial samples drawn were designated as upper and the others from the lower portion were designated as lower. Two photomicrographs were taken of each sample. Results obtained from these tests are summarized in Table 4.3. Sample droplet size distributions are presented in Figures 4.10.

Group average droplet diameters of 2.8 J.lm and 3.1 J.lm based on dN,L were computed for emulsions 1 and 2 respectively. These group average droplet diameters are quite similar. Therefore, based on these results, it can be concluded that the preparation method developed in this study produced emulsions with reproducible droplet size distributions for emulsions of similar composition

Group average droplet diameters of 3.2 and 2.3 J.Lm based on dN,L were computed for the upper and lower portions of Emulsion 1 respectively. Group average droplet diameters of 2. 7 and 3.4 J.Lm based on dN,L were also computed for the upper and lower portions of Emulsion 2 respectively.

64 These group average droplet diameters are greater than the group average droplet diameter of 1. 7 J.lm computed for samples A to D for the 25 wt% emulsion shown previously in Table 4.2. This discrepancy might be related to the difference in water concentrations. Group standard deviations of 0.6 J.lm and 0.5 J.lm were computed for emulsions 1 and 2.

The results from Section 4.2.1 and 4.2.2 prove the stability and reproducibility of the emulsions utilized in this study.

65 Table 4.3 Mean droplet diameter and standard deviation data for 30 wt% water-in-N-Brightstock oil emulsion reproducibilty tests.

Mean Sauter Sample Group Group Point of Number droplet mean Standard average standard sample of droplets diameter diameter deviation based on deviation Emulsion Sample collection counted dN.L_fu_m) ds.v (p.m) (p.m) dN.L{Jlm) (p.m) Emulsion 1 Ia UW_er 215 2.9 19.7 4.1 lb U_pper 283 2.9 15.9 3.3 3.2 2a Upper 185 3.2 20.6 4.1 I 2b Upper 151 3.8 19.6 4.3 3a Lower 240 2.1 16.6 3.1 3b Lower 223 2.1 28.5 4.1 2.3 4a Lower 239 2.5 47.1 5.2 0\ 0\ 4b Lower 270 2.5 10.0 2.3 2.8 0.6 Emulsion 2 Ia Upper 291 2.2 24.2 3.8 lb Upper 232 2.5 7.5 2.1 2.7 2a Upper 257 3.1 32.2 4.9

! 2b Upper 325 3.0 44.4 6.0 3a Lower 218 3.3 31.0 4.9 3b Lower 283 3.0 20.8 4.1 3.4 4a Lower 244 3.7 20.6 4.2 4b Lower 209 3.7 23.3 4.7 3.1 0.5 -~~~~~- - ~ ~00 CD -~f!BO -aCD § 70 >. u ;oo -+- EmJsion 1 SarrPe 1a -11- EmJsion 2 SarrPe 2a o-::::s ....f 50 CD ~40 .!! ::::s E30 ::::s u CD2() ;::> -i 10 0::

0+-----~~~~------~------~------~------~ -0.5 0 0.5 1 1.5 2 logarithm of droplet clameter (1-111)

Figure 4.10 Droplet size distributions of 30 ~/o water-in-N-Brightstock oil Emulsion 1, Sample 1a and Emulsion 2, Sample 2a.

67 4.3 Self-Lubricating Flow in a Couette Cell The concentric cylinder viscometers (Haake RV3 and Haake RS150) were used to study the self-lubricating flow of viscous oils. The oils were sheared in the annular gap between the rotating spindle and the cup. The magnitude of the torque required to rotate the spindle was recorded as a function of time. This was an indication of the state of the sheared fluid. The experimental parameters investigated in this research are summarized in Table 4.4. Details of all experimental runs conducted in this research are shown in Table B.1 in Appendix B.

Table 4.4 Summary of experimental parameters studied during the investigation of self-lubricating flow in a Couette cell.

Oil N-Brightstock, Shellflex 810, Bitumen

Water Concentration (wto/o) 10 ' 17 ' 25 ' 30 ' 35 Temperature ~C) 15 ' 22.5 ' 30 ' 40 ' 50 Spindle Speed (rpm) 8 to 512 Radius Gap width between spindle and Spindle Type (mm) cup (mm) MVIII (Haake RV3) 15.200 5.800 Z31 Ti (Haake RS 150) 15.725 5.942 Z38 Ti (Haake RS150) 19.010 2.657 Z41 Ti (Haake RS 150) 20.710 0.957

It should be noted that solids were not added to the emulsions in these study. Joseph et al. (1999) suggest a mechanism for self-lubricating flow of bitumen froth based on the dispersed clay particles present. The results obtained in this study suggest that the presence of solids may not be critical to self-lubricating flow.

68 There were two typical types of tests conducted. One involved the measurement of torque versus time at constant spindle speed and temperature; the other involved stepwise incremental changes in spindle speed at a constant temperature.

The constant spindle speed procedure described in Section 3.4.1 involved shearing the emulsion at a single spindle speed and constant temperature for a duration of about 1 hour. The step-wise spindle speed procedure involved shearing the oil at different spindle speeds for durations ranging from 5 to 10 min determined to be a reasonable enough time to ensure that there is no change in measured torque. For a particular temperature, the step-wise spindle speed procedure preceded the constant spindle speed procedure and was normally used as a quick method of obtaining an estimate of the (critical) spindle speed at which self-lubricating flow first occurs. Subsequently, the constant spindle speed procedure was then used at each spindle speed to confirm the results obtained from the step-wise spindle speed procedure.

4.3.1 Development of a Criterion for Assigning Self- Lubricating Flow Three different flow patterns were identified based on experimental observations. i) Self-Lubricating Flow (An example is illustrated in Figure 4.11) ii) Intermediate or Marginal Lubricating Flow (An example is illustrated in Figure 4.12). iii) Viscous (Non-Lubricating) Flow. An example is illustrated in Figure 4.13

69 20000~------~ ..-.. E z ..._.:::L ~15000~------~ ::::s e-o l- 10000~------~

0 500 1000 1500 2000 2500 3000 3500 4000 Time (sec)

Figure 4.11 Self-Lubricating Flow of 30wt% water-in-N-Brightstock oil emulsion.

Run no. 105; T = 30°C, Z31 spindle (R1 = 15.725 mm), n = 181 rpm. Ratio of initial to final torque is approximately 11.

70 34300 8 z 24500 ...__,::t Cl) 14700 ~ 0 4900 ~

1200 2400

I Time (s)

Figure 4.12 Marginal Self-Lubricating Flow of 30 wt% water-in-N-Brightstock oil

emulsion. Run no. 119; T = 30°C, MVIII spindle (R1 = 15.2 mm), n = 64 rpm. Ratio of initial to final torque is 7.5.

71 20000.------~ 20000r~~~~~=-~==~-====~-====~l

-. zE1oooo~------~ .._.::s. ::sCD ...0" 010000~------~ 1-

5000+------~

0+------r------~----~----~------~----~.~----,-----~ 0 500 1000 1500 2000 2500 3000 3500 4000 Time (sec)

Figure 4.13 Viscous (Non-Lubricating) Flow of 30wt% water-in-N-Brightstock oil

emulsion. Run no. 104; T = 15°C, Z31 spindle (R1 = 15.725 mm), n = 8 rpm. Ratio of initial to final torque is approximately 1.2.

72 The fact that a constant measured torque should be recorded with respect to time for a Newtonian fluid has been discussed in Chapter 2. It can be observed from Figures 4.11 to 4.13 that the measured torque varies with time for the emulsions of this study. The observations shown in Figures 4.11 to 4.13 therefore do not indicate that the emulsions exhibit Newtonian fluid behaviour (although N-Brightstock oil has previously been shown to be Newtonian within a certain temperature range).

Figure 4.11 indicates an order of magnitude variation in the initially high torque. The initial torque : final torque ratio is about 11. The resulting low torque is believed to coincide with the formation of a lubricating water layer.

Figure 4.12 indicates more reduction in torque than Figure 4.13 but the reduction is not as significant as in Figure 4.11. The torque reduction ratio in Figure 4.12 is 7.5. This is an exceptional case and the basis for the marginal self-lubricating flow will be explained. 5 out of a total of 218 experimental points resulted in marginal self­ lubricating flow.

Figure 4.13 indicates little or no variation in torque. The initial torque reduction ratio is about 1.2.

A criterion was developed based on the experimental results obtained in this research to distinguish between the three flows observed. The ratio of the initial to the final torque was used to distinguish between the three different types of flow behaviour.

Experiments where the torque reduction was less than 2 have been designated as 'Non­ lubricating flow'. Experiments where torque reductions were between 2 and 5 have been designated as 'Marginal lubricating flow'. Experiments where the torque reduction ratio was greater than 5 have been designated as 'Self-lubricating flow'.

Each experimental point in the tests involving the stepwise spindle speed procedure is considered as a single point, for example, Runs 73a and 73b are designated as two

73 experimental points while Run 74 is designated as a single point. The total number of experimental points was 218. This comprises 62 self-lubricating flow, 7 marginal lubricating flow, and 149 non-lubricating flow experimental points.

The criterion for classifying flow types was valid for 53 out of the 62 self-lubricating flow experimental points obtained. The remaining 9 experimental points were mainly results of the stepwise spindle speed tests in which the shear history affects the final torque values. Ratios from the stepwise spindle tests were normally found to be lower than the constant spindle tests at the same conditions.

The criterion was valid for 5 out of the 7 marginal lubricating flow points obtained. It should be noted that the marginal lubricating flow shown in Figure 4.12 is one of the exceptions since it has a reduction ratio (7 .5) greater than 5 and was only classified visually from the plot as marginal self-lubricating flow.

The criterion was valid for 148 out of the 149 non-lubricating flow points obtained.

4.3.2 Reproducibility of Experimental Runs The reproducibility of the experimental results is summarized in Table 4.5. The number of times the same result was obtained represents the measure of reproducibility. It can be observed from Table 4.5 that the experiments are reproducible.

A detailed analysis of reproducibility is provided in Table B.2 in Appendix B. It can be observed from Table B.2 that consistent results were obtained with the repetitions with the exception of two runs. Two out of a total of three repetitions conducted at 64 rpm and 30°C resulted in Marginal lubricating flow. Similarly, two out of a total of three repetitions conducted at 128 rpm and 40°C resulted in Marginal lubricating flow. The reasons for the deviating results obtained from runs 75d and 84f are not known.

74 Table 4.5 Measure of reproducibility of Couette flow experimental results for 30 wt% water-in-N-Brightstock oil emulsion. Abbreviations SLF, NLF and MLF represent Self-lubricating flow, Non-lubricating flow and Marginal lubricating flow respectively.

Spindle No of times No of times I speed Type of result result not Temperature c-c) (rpm) Method Lubricating flow No. of repetitions obtained obtained 30 90.5 Stepwise SLF 1 1 0 30 90.5 Constant SLF 5 5 0 40 181 Stepwise SLF 1 1 0 -.) 40 181 Constant SLF 2 2 0 Ul 50 256 Stepwise NLF 1 1 0 50 256 Constant NLF 3 3 0 50 362 Stepwise SLF 1 1 0 50 362 Constant SLF 5 5 0 50 512 Constant SLF 2 2 0 The conditions at which the greatest number of replicate runs were conducted using the 30 wt% water-in-N-Brightstock oil emulsion are shown in Table 4.6. The average of the initial and fmal torques for each experimental condition are shown. The 256 rpm run resulted in non-lubricating flow. The 90 rpm and 362 rpm runs resulted in self­ lubricating flow. The standard deviation of the torques appear to be of a magnitude comparable to the torques.

Table 4.6 Comparison of initial and final torques obtained from Couette flow experiments with 30 wt% water-in-N-Brightstock oil emulsion at 30 and 50°C.

S_p_indle speed (rpm) 90.5 256 362 Temperature tC) 30 50 50 Number of repeats 6 4 5 Average of initial torques (N m) 0.033 0.017 0.019 Average of fmal torques (Nm) 0.0029 0.013 0.0040 Standard deviation of initial torques (Nm) 0.0032 0.0026 0.0042 Standard deviation of fmal torques (Nm) 0.0014 0.0019 0.0029

4.3.3 Comparison of Observed Initial Torgue With Theoretical Predictions The average of initial torques from replicate runs are compared with the theoretically predicted torques (using equation 2.1) based on the viscosity of pure N-Brightstock in Table 4.7. It can be observed that the predicted torque is less than the averaged initial torques. This agrees with the results of previous research on emulsions in which the viscosity of water-in-oil emulsions is expected to be greater than that of the pure oil (Benayoune et al., 1998).

76 Table 4. 7 Comparison of observed initial torques obtained during self-lubricating flow of 30 wt% water-in-N-Brightstock emulsion with theoretical predictions. Average of Initial Torques Spindle from Predicted Temperature speed experiments Torque % (»C) (rpm) (Nm) (Nm) Difference 22.5 22.6 0.021 0.013 65.2 22.5 32 0.026 0.018 45.2 30 90.5 0.033 0.023 44.1 40 181 0.024 0.017 44.6 50 256 0.017 0.010 70.6 50 362 0.019 0.014 29.5 50 512 0.027 0.020 30.6

4.4 Development of Self-Lubricating Flow Maps Previous research by Charles et al. (1961) has shown that the concentric oil-in-water flow pattern in core-annular flow is only obtained under a range of conditions. Charles et al. ( 1961) represented their results as superficial oil velocity versus superficial water velocity plots, pressure gradient versus input oil-water ratio plots amongst others. These plots had boundaries demarcating the different flow regimes obtained. Joseph et al. (1997) presented a similar plot for flow in a vertical pipeline. The experimental results obtained in this research are presented as a series of temperature versus spindle speed plots.

4.4.1 N-Brightstock Emulsions The experimental results obtained by shearing 30 wt% water-in-N-Brightstock oil emulsion in the concentric cylinder viscometer are presented in Table B.3 in Appendix B and plotted in Figure 4.14.

The following observations can be drawn from Figure 4.14:

77 i) There appears to be a region which represents a transition between Viscous (Non­ lubricating) flow and Self-Lubricating flow. ii) At a given constant temperature, achieving Self-Lubricating flow is a function of spindle speed (shear rate). iii) The critical spindle speeds required to achieve Self-Lubricating flow increases with increasing temperature.

60

50 ...... A A ...... ~ ~

...... !Nil A A A A ...... ~ ~ ...... - ~ ~

...... A A A ...... -Jlll\ll ~ ~ ~

-~~ ~

~~ • Non - Lubricating flow 10 I-- 11 Marginal Lubricating flow A Self - Lubricating flow

0 0 100 200 ~ ~0 500 600 Spindle speed (rpm)

Figure 4.14 Flow map for 30 wt% water-in-N-Brightstock oil emulsion (Temperature and spindle speed as independent variables).

78 Self-Lubricating flow and the Non-Lubricating flow patterns shown in Figure 4.14 may be compared to the concentric oil-in-water (core-annular flow) and water droplets-in-oil flow patterns obtained by Charles et al. (1961). Charles et al. (1961) discovered two boundaries enclosing the core-annular flow pattern; in other words, a lower and an upper critical velocity. The lower critical velocity represents the velocity at which the core-annular flow regime is attained. The upper critical velocity represents the velocity at which there is a transition from the core-annular flow pattern to the oil slugs-in-water flow pattern. Joseph et al. (1999) reiterates this fact when he stated that a lower critical velocity was discovered for self-lubricating flow but an upper critical velocity, which was believed to exist, was not discovered. Kruka (1977) also found a lower critical velocity for self-lubricating flow in his experiments.

Results obtained in this study agree with those ofKruka (1977) and Joseph et al. (1999) in that no upper critical velocity was discovered for self-lubricating flow. It is possible that at a reasonably high spindle speed, re-emulsification and subsequent destruction of the self-lubricating flow pattern will occur in agreement with Charles et al. (1961).

The gradual change in flow patterns was best observed during the stepwise spindle speed procedure. As the critical speed was approached, elongated droplets were observed at the exposed upper emulsion surface (this might be an indication of Intermediate/Marginal flow). These elongated droplets were observed to increase in length as the spindle speed was increased. The elongated droplets eventually coalesced to form what appeared to be a lubricating water layer at the critical spindle speed.

The requirement of a critical shear rate for self-lubricating flow suggests that there may in fact be a critical shear stress required to attain self-lubricating flow. The experimental data shown in Figure 4.14 can be re-plotted as shown in Figure 4.15. Figure 4.15 appears to indicate the requirement of a critical shear stress (equivalent to torque) for self-lubricating flow. Self-lubricating flow appears to occur above an upper torque limit.

79 Results from Figure 4.14 demonstrated that the critical speed required to achieve self­ lubricating flow was observed to increase with increasing temperature. Joseph et al. (1999) discovered that the critical velocity required to lose self-lubricating flow decreases with increasing temperature. Joseph et al. (1999) believed that the critical velocity required to lose self-lubricating flow was close to the critical velocity required to achieve self-lubricating flow. The procedure used by Joseph et al. (1999) to obtain the critical velocity involved decreasing the velocity in steps at a particular temperature and measuring the pressure gradient.

50000.------.

45000

40000

35000 6. E 30000 • -z ~ -Cl) 25000 6 ~ C" ~ ....0 20000 •• • • • 15000 •• • • Non - Lubricating flow 10000 • • • • • 11 Marginal self- lubricating flow • • ll Self- lubricating flow 5000 ••••

0+------~~------~------~------r------.------~ 0 100 200 300 400 500 600 Spindle speed (rpm)

Figure 4.15 Flow Map for 30 wt% water-in-N-Brightstock oil emulsion (Measured torque and spindle speed as independent variables).

80 Two tests were conducted with 30 wt% water-in-N-Brightstock emulsions at 22.5 and 50°C using a modified form of the step-wise spindle speed procedure to determine the critical velocity required to lose self-lubricating flow. At 22.5 and 50°C, spindle speeds higher than critical (45.2 rpm and 512 rpm respectively) were employed as starting spindle speeds. After self-lubricating flow was maintained, the spindle speeds were decreased gradually. This is the opposite of the step-wise spindle speed procedure previously described.

The critical speed for decreasing spindle speeds was defined by Sanders et al. (2004) as the lowest speed at which the low torque (for self-lubricating flow) was measured. Applying this definition to the emulsion tests described above, the critical spindle speeds for the loss of self-lubricating flow were thus determined as 32 rpm at 22.5°C and 362 rpm at 50°C. This agrees with previous results obtained using the increasing stepwise spindle speed procedure in which self-lubricating flow was achieved at critical spindle speeds of 32 rpm at 22.5°C and 362 rpm at 50°C respectively. The results from these two limited tests appear to show that the method used with the stepwise-spindle procedure (decreasing or increasing) does not affect the critical spindle speeds. These results do not agree with those obtained by Sanders et al. (2004) in which the critical speeds required for the self-lubricating flow of bitumen froth were observed to change when the decreasing step-wise approach was employed.

The difference between the trends in the results obtained in this study and those of Sanders et al. (2004) and Joseph et al. (1999) might be due to the temperature variation between the bitumen being sheared and the fluid in the heating device (Sanders, 2003).

Results obtained from experiments involving shearing bitumen froth in the Haake RV3 concentric cylinder viscometer by Sumner et al. (2003) are replotted as shown in Figure 4.16 and also provided in Table B.4 in Appendix B. It should be noted that reproducibility of the bitumen froth experiments was not verified as some repetitions

81 produced varying results. As previously mentioned, the bitumen froth had between 15 to 30 wt% water.

70

A A A 60 ~ ~ u

A A 50 ~ "-' -(J .6.6. 6. 0

~AA A -!40 ~-~ .....:I I! ~ CD 0.30 E ~ 20 • Non - Lubricating flow a Marginal Lubricating flow !!:. Self - Lubricating flow 10 ~

0 0 100 200 300 400 500 600 700 800 Spindle speed (rpm)

Figure 4.16 Flow Map for water-in-Bitumen emulsion (Temperature and spindle speed as independent variables).

82 Within the limit of the experimental conditions tested, it can be observed from Figure 4.16 and Table B.4 that the critical spindle speed required for self-lubricating flow of bitumen emulsions appears to increase with increasing temperature. A transition region between self-lubricating flow and non-lubricating flow also appears to be present. These observations are similar to the observations made for the 30wt% N-Brightstock emulsion in Figure 4.14.

It may thus be concluded that the N-Brightstock emulsion exhibits behaviour similar to bitumen froth in a concentric cylinder geometry.

4.4.2 Other Emulsions

Limited shearing experiments with a 30 wt% water-in-Shellflex 810 oil emulsion also resulted in observations similar to that made for the 30 wt% water-in-N-Brightstock emulsion previously shown in Figure 4.14. The results obtained for the 30 wt% water­ in-Shellflex 810 oil emulsion are presented in Figure 4.17. It can be observed from Figure 4.17 that the critical spindle speed increases with increasing temperature. Achieving self-lubricating flow can also be observed from Figure 4.17 to be a function of shear rate. Self-lubricating flow was not obtained at the higher temperature of 50°C.

A 30 wt% water-in-Bitumen emulsion was also sheared at spindle speeds ranging from 16 to 32 rpm at 50°C. Self-lubricating flow was observed at 32 rpm.

83 60

...... 50 ,.. ,...... ,...... ,.,......

40 0 ~ ...... A !30 ,.. ,.. ,...... ,.,...... L.J. ...:::::s ~ CD C.20 E CD 1- 10

+ Non - lubricating flow ..-..-...... _.... A A A 0 ...... L.J. L.J. L.J. I I 6. Self-lubricating flow ~

-10 I 0 100 200 300 400 500 600 700 BOO Spindle speed (rpm)

Figure 4.17 Flow Map for 30 wt% water-in-Shellflex 810 oil emulsion (Temperature and spindle speed as parameters).

4.5. Effect of Shear on Water Droplet Size Examples of photomicrographs of the N-Brightstock emulsion taken before and after shear for an experiment which resulted in self-lubricating flow are shown in Figures 4.18a and 4.18b respectively. The corresponding droplet size distributions are compared in Figure 4.19.

Figure 4.18a is a photomicrograph of the emulsion before shear while Figure 4.18b is a photomicrograph of the emulsion after shear. Figures 4.18a and 4.18b show non­ spherical water droplets in an oil matrix. The arrow in Figure 4.18b indicates what is likely an air bubble at the bottom right comer which was not included in the droplet

84 sizing. Relatively fewer water droplets can be observed in Figure 4.18b. An arithmetic mean diameter and Sauter mean diameter of 2.8 and 15.4 J.lm respectively was computed for the droplets in Figure 4.18a. An arithmetic mean diameter and Sauter mean diameter of5.2 and 33.7 J.lm was computed for the droplets in Figure 4.18b which also indicates a certain degree of coalescence with shear.

It can be observed from Figure 4.19 that the droplet size distributions before and after shear are not exactly the same. It appears that there is an increase in the droplet size after shear.

85 (b) Figure 4.18 Photomicrographs showing water droplets in oil matrix for Self- lubricating flow case. (a) before shear and (b) after shear. Arrow in (b) indicates a probable air bubble.

86 100 ...... ~90 CD N -~ 80 CD "C § 70 >.u c 60 CD -+-Droplet diameter data before shear (Run 101-2) ::I cr ...! 50 -a-Droplet diameter data after shear (Run 101-4) CD > ~40m "3 E 30 ::Iu ~ 20 ~m &10

0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 Logarithm of droplet diameter (pm)

Figure 4.19 Droplet size distribution for 30 wt<»/o water-in-N-Brightstock oil emulsion before and after shearing with concentric cylinder viscometer for Self-Lubricating flow.

Run 101; T =l5°C, n = 32 rpm, Z31 spindle (R1 = 15.725 mm).

87 Photomicrographs for an experiment which resulted in non-lubricating flow before and after shear are presented in Figures 4.20a and 4.20b. The corresponding droplet size distributions are shown in Figure 4.21. An arithmetic mean diameter and Sauter mean diameter of 3.0 and 38.9 J..Lm were computed respectively for the droplets in Figure 4.20a. An arithmetic mean diameter and Sauter mean diameter of3.2 and 31.9 J..Lm was computed for the droplets in Figure 4.20b. In this case, the mean diameters do not indicate a clear increase in droplet diameter with shear.

It can also be observed from Figure 4.2 that there is no consistent change in droplet size distributions before and after shear. The degree of variation between the curves also appears to be of a smaller magnitude than that observed previously for the self­ lubricating flow case in Figure 4.19. This would also seem to suggest that there is not a significant change in droplet size with shear when the flow does not exhibit non­ lubricating flow.

88 (a)

(b) Figure 4.20 Photomicrographs showing water droplets in oil matrix for Non-lubricating flow case. (a) before shear and (b) after shear. Arrows in (b) indicates probable air bubble.

89 100

-?!. 90 -CD N "! 80 CD "ts c 70 :l ~ -+-Droplet diameter data before shear (Run 95-2) g 60 CD -a-Droplet diameter after shear (Run 95-3) :l ! 50 """CD .2!:..., 40 .!! :l E 30 :lu ~ 20 ..as &. 10

0 0 0.5 1 1.5 2 Logarithm of droplet diameter (pm)

Figure 4.21 Droplet size distribution for 30 wt% water-in-N-Brightstock oil emulsion before and after shearing with concentric cylinder viscometer for Non-Lubricating flow.

Run 95; T = 50°C, n = 90.5 rpm, Z31 spindle (R1 = 15.725 mm).

4.6 Effect of Water Concentration on Self- Lubricating Flow The flow map for a 30 wt% water-in-N-Brightstock oil emulsion was shown in Figure 4.14. Experiments were also conducted with water-in-N- Brightstock oil emulsions in which water concentration was varied from 10 to 35 wt%. An attempt to produce an emulsion with 45 wt% water was unsuccessful. The oil and water phases could not be completely dispersed by the homogeniser.

90 Detailed results are provided in Table B.S in Appendix B. Flow maps for each water concentration are provided in Figures B.l to B.4 in Appendix B. The plots all appear to agree with the observations previously noted for 30 wt% water-in-N-Brightstock oil emulsion in Figure 4.14 within the limits of the experimental data provided. The critical speeds are observed to increase with increasing temperature. A transition region between non-lubricating flow and self-lubricating flow also appears to be present. It can also be observed from Figures B.1 to B.4 within experimental limits that at the higher temperature of 50°C, the difficulty of achieving self-lubricating flow increases with decreasing water concentration.

Anomalous behaviour can be observed from the 10 wt% N-Brightstock emulsion flow map in Figure B.l At the lower temperature of 22.5°C and a spindle speed of 362 rpm, the previously attained self-lubricating flow was lost. This experiment was conducted using the stepwise spindle speed procedure previously described. The preceding experimental points at 181 rpm and 256 rpm were repeated once using the constant spindle speed procedure and the results were found to agree with the results previously obtained using the stepwise spindle speed procedure. The reason for this anomalous behaviour is not known.

Figure 4.22 shows a plot of the critical spindle speeds required to achieve self- · lubricating flow against water concentration with temperature as a parameter.

91 40

+22.5°C 35 ...... - - •30°C 650°C -30 ...... ~ - ! - - c 25 ...... 0 - ;:; - - l! 1:20 CD (,) c 0 A (,) 15 • ~ CD ...cu ~ 10 ......

5

0 0 100 200 300 400 500 600 700 800 Critical spindle speed (rpm)

Figure 4.22 N-Brightstock oil emulsion water concentration as a function of the critical spindle speed required to achieve self-lubricating flow with temperature as a parameter.

The following observations can be made from Figure 4.22:

i) The critical spindle speed required to achieve self-lubricating flow decreases with increasing water concentration. This is believed to be due to the presence of a greater amount of water which is available to form the water layer. ii) At a particular water concentration, the critical spindle speed increases with increasing temperature. This agrees with previous observations made for 30 wt% water in N-Brightstock oil emulsion from Figure 4.14.

92 iii) There appears to be an optimum water concentration of 30 wt%. There is no change in critical spindle speed as the water concentration is increased to 35 wt%.

4. 7 Effect of Shear History on Self-Lubricating Flow

The initial torque reduction time in this study is defined as the time required for the initially observed high torque to reduce to the lower torque associated with the formation of the lubricating water layer. This is illustrated in Figure 4.23. Initial torque reduction time was studied with the 30 wt% water-in-N-Brightstock oil emulsions but the observed behaviour is believed to be applicable to other emulsion concentrations. Results are shown in Table B.6 in Appendix B and Figure 4.24.

Initial torque Persistent self lubricating behaviour reduction time

Time

Figure 4.23 Schematic representation of torque reduction time for self-lubricating flow.

93 It can be observed from Figure 4.24 that the initial torque reduction time generally decreases with increasing spindle speed at a constant temperature for the lowest temperatures studied (15, 22.5°C}. At higher temperatures, there does not appear to be an appreciable variation in torque reduction time with spindle speed at a constant temperature.

2500

•• ---15°C u2ooo .. -..--22.5°C CD U) _._30°C I ·~I -CD -e-4ooc I E I -·•-·50°C I = I s 1500 I I CJ I = I ,:J ! I 'I 1000 ! I e- II \ .9 I I Ci I :2 I .5 500 \ I ~· ..L .. -- _c., ""=" - - <.::.7 -~ - I 0 0 100 200 300 400 500 600 Spindle speed {rpm)

Figure 4.24 Initial torque reduction time as a function of spindle speed with temperature as a parameter for self-lubricating flow of 30 wt% water-in-N-Brightstock oil emulsions.

94 The lowest initial torque reduction durations appear to be obtained at the higher temperatures (and high spindle speeds). The initial torque reduction period shown in Figure 4.23 corresponds visually to a sort of disturbance in the emulsion. This will be explained in Section 4.1 0. A shorter period of torque reduction indicates a more rapid progression to the self-lubricating flow regime.

4.8 Effect of Spindle Size (Gap Width) on SeH-Lubricating Flow

The objective of these tests was to determine if self-lubricating flow would occur when 30 wt% N-Brightstock emulsions was sheared with the medium diameter Z38 and large diameter Z41 spindles as in the case with the smaller Z31 spindles. Sumner et al. (2003) had previously been unable to achieve self-lubricating flow with medium and large diameter spindles.

The results obtained from shearing 30 wt % N-Brightstock emulsions with the three available spindles Z31, Z38 and Z41 is shown in Figure 4.25.

It can be observed from Figure 4.25 that the shear rates at the surface of the smallest spindle (Z31) used in this study are comparable to those of the larger spindles. Shearing with the medium sized spindle (Z38) resulted in self-lubricating flow while shearing with the large spindle (Z41) did not result in self-lubricating flow. This is in agreement with results obtained by Sumner et al. (2003) for oil emulsions. Self-lubricating flow of N-Brightstock emulsions was obtained in this study by shearing with the medium sized spindle (Z38). Self-lubricating flow of emulsions with medium and large spindles was not obtained by Sumner et al. (2003). The inability to achieve self-lubricating flow with the medium and large spindles by Sumner et al. (2003) may be caused by the low applied shear stresses (torques) available on the Haake RV3 viscometer used.

95 40

411M6 lI •-ana-..._...lul>ricllllng_ "~ana-

10

50 100 150 200 250 300 350 Shear rate lti!Pindle surf- (s"') 45

40

35 e.-30

125 ... 6 6 6 120 i 4-----.-·~- .... ,5 I

10

50 100 150 200 250 300 350 Shear r•at spindle eurfKe <•')

45

40

35 e.-30 ~25 ! 120 • • • E l ·-ana- t! 15

10

50 100 150 200 250 300 350 Shear rate II -.Nnclle eurfKe <•')

Figure 4.25 Effect of spindle size on self-lubricating flow of 30 wt% N-Brightstock emulsion. Flow maps obtained from shearing with Z31, Z3 8 and Z41 spindles respectively.

96 The results obtained in this research suggest that some degree of radial shear rate variation is required for self-lubricating flow. In single phase Couette flow using the large spindle, the shear rate changes by less than 10% from the spindle to the cup as described in Appendix E. This is in contrast to the 30% and 90% reductions corresponding to the use of the medium and small spindles, respectively.

However, it is also possible that extending the shear rate range used with the large spindle to extremely high shear rates might also cause self-lubricating flow.

The experiments also demonstrated that the critical spindle speeds required to achieve self-lubricating flow using the small and medium sized spindles are quite similar. Both spindles required critical speeds of 32 rpm at 22.5°C and critical speeds of 181 and 256 rpm, respectively, at 40°C. It should be noted that the shear rates are different at the same speed for the two spindles because they differ in radii.

A comparison can be made between the wall shear rate in the pipeline experiments by Joseph et al. (1999) and the spindle shear rates used in this research. The wall shear rates in the pipeline (8V /D) computed at a velocity of 1.5 m/s for pipe diameters of 25 1 mm, 50 mm and 0.6 mm are approximately 472, 236 and 20 s- • The spindle shear rates from Figure 4.25 appear to be in the range of these pipeline wall shear rates. This similarity in shear rates strengthens the justification for comparing results obtained by Joseph et al. (1999) with the results obtained in this research.

4.9 Effect of Tvoe of Oil On Self-Lubricating Flow

As previously observed with the results from shearing N-Brightstock emulsions, shear rate is a very important factor in achieving self-lubricating flow. It was thus expected that provided enough shear was applied to an emulsion of any viscous oil, self­ lubricated flow should be achieved. In other words, self-lubricated flow was expected to be independent of the type of oil.

97 Shearing of Shellflex 810 oil emulsions was conducted at low temperatures in an attempt to obtain a viscosity similar to that ofN-Brightstock at 22.5°C. The viscosity of Shellflex 810 oil at 0°C was determined to be 13.63 Pa.s . A comparable viscosity of 14.47 Pa.s was obtained with N-Brightstock oil at 22.5°C. Shellflex 810 oil and water were cooled to 0°C prior to preparing a 30 wt% water-in-Shellflex oil emulsion to be sheared. The stepwise spindle speed procedure was employed to find the critical spindle speed required to achieve self-lubricated flow. The critical spindle speed for the Shellflex emulsion at 0°C was determined to be 64 rpm. Subsequent repetitions were conducted using the constant spindle speed procedure to confirm the critical spindle speed. These repetitions confirmed the previous result. A Shellflex oil emulsion was also sheared at 50°C for comparison with the results obtained with the N-Brightstock emulsion. As previously noted in Section 4.4.2, self-lubricating flow was not obtained at 50°C.

Bitumen was also preheated to 50°C and sheared with water to produce a 30 wt% water-in-bitumen emulsion. The stepwise spindle speed procedure was also employed at 50°C with bitumen to find the critical speed. A critical spindle speed of 32 rpm was determined for the bitumen emulsion at 50°C.

Results from these experiments to determine the effect of oil viscosity on self­ lubricating flow are summarized in Table B.1 0 in Appendix B and plotted in Figure 4.26. A critical spindle speed of 32 rpm obtained for the Shellflex oil emulsion was only one spindle speed adjustment lower than that for N-Brightstock (64 rpm) when their oil viscosities were 13.63 and 14.71 Pa· s respectively. The critical spindle speed of32 rpm obtained for the N-Brightstock emulsion was one spindle speed adjustment higher than that for the Bitumen emulsion (22.6 rpm) when the oil viscosities were similar (32.51 and 25.63 Pa· s respectively). This would suggest that the type of oil used for the emulsion may not be as important as the oil viscosity for the limited number of oils considered in this study.

98 The ratio of N-Brightstock oil viscosity to water viscosity for the temperature range 10 to 50°C is computed to be between 49684 and 1894. Similar ratios are computed for Shellflex as 17210 and 493 between 0 and 50°C and 46645 for bitumen at 50°C. Oliemans and Ooms (1986) state that core-annular flow will be obtained by crude oils 3 with viscosity greater than 500 mPa· s and densities greater than 900 kglm • Bannwart (200 1) states that crude oils with viscosities greater than 100 mPa· s and densities close to that of water will attain core-annular flow.

Joseph (1998) states that core-annular flow of heavy oil will occur with water at the pipe wall if the oil viscosity is larger than 500 mPa.s. He also states that drag reductions of the order of the viscosity ratio are possible with the ratio of the viscosity of oil to the 5 viscosity of water equal to 10 •

35

• • N Brightstock 30 -· •Shellflex 810 •Bitumen 25 •

• •

10

5

0 I I • I • 0 100 200 300 400 500 600 700 800 Critical spindle speed (rpm) Figure 4.26 Oil viscosity as a function of critical spindle speed required to achieve self-lubricating flow (30 wt% water-in-oil emulsions).

99 The results obtained in this study satisfy all the above criteria for core-annular flow. N­ Brightstock, Shell flex 810 and Bitumen had viscosities greater than 500 mPa· s and 3 densities greater than 900 kglm • The viscosity ratios obtained in this study are of the 2 5 order of 10 to 10 •

4.10 Visual Observations

The flow structure formed during the experimental tests could only be observed at the free surface of the emulsion in the viscometer. The flow structure along the length of the spindle could not be observed. The progressive change in the structure resulting from stepwise increases in spindle speed will be discussed for the non-lubricating flow case and the self-lubricating flow case. The non-lubricating flow case is illustrated in Figure 4.27 while the self-lubricating flow case is illustrated in Figure 4.28.

At the start of the experiment, the surface of the emulsion was level with the top of the spindle. At speeds lower than the critical speed required to achieve self-lubricating flow, elongated droplets were observed to orbit in a defined circular path at a radial position of about 0.4 to 0.5 of the gap width. This is illustrated in Figure 4.27c. Similar observations were made for non-lubricating flow tests using the constant spindle speed procedure.

As the critical speed required to achieve self-lubricating flow is approached, a dip in the level of the fluid occurs. A convex shaped emulsion surface was observed around the spindle as illustrated in Figure 4.29. A concave meniscus was observed on the emulsion surface from the cup to the water layer as shown in Figure 4.29.

100 Emulsion Elongated droplets Spindle Emulsion B

Orbital path ...... 0 ...... Emulsion

Cup ( c ) ( a ) ( b )

Figure 4.27 Schematic diagram (Plan view) of progression from original emulsion to Non-Lubricating Flow. (a) Original emulsion (b) Boundary B (shown in Figure 4.29) observed between emulsion layers (Not observed in all cases). (c) Elongated water droplets observed orbiting in a defined path. Boundary B as shown in Figure 4.29 is the common boundary between the convex shaped emulsion surface and the water layer. The boundary B shown in Figure 4.28b is prominent and is always observed for the self-lubricating flow case. At the critical speed for self-lubricating flow, a disturbance was observed in the emulsion. The disturbance consisted of oscillatory movements in the concave meniscus. The duration of this disturbance corresponds to the initial torque reduction period previously discussed in Section 4.7 and illustrated in Figure 4.25.

The disturbance eventually disappeared and a continuous water layer was observed around the emulsion layer· surrounding the spindle. The lubricating water layer was normally observed between 0.3 to 0.6 of the gap width between the cup and the spindle. The thickness of the water layer was visually estimated to be of the order of 1 mm. The model developed in Chapter 5 provides an estimate of the water layer thickness. This visually estimated water layer thickness is less than the 2 mm estimated for flow in a 50 mm pipeline by Neiman (1986) and is greater than the range 0.3- 0.4 mm estimated by Joseph et al. (1999) for flow in a 25 mm pipeline.

Some droplets were observed on the outside of the boundary B shown in Figure 4.28b. These droplets are much smaller than the elongated droplets discussed previously for the non-lubricating flow mechanism.

The emulsion layer with the convex surface around the spindle that occurs with self­ lubricating flow is well defined at the low temperatures (22.5°C and 30°C) and is less prominent at the higher temperatures (50°C).

102 Emulsion Lubricating Water layer Spindle Emulsion Emulsion

Emulsion

...... 0 w Emulsion

Cup

( a ) ( b ) ( c )

Figure 4.28 Schematic diagram (Plan view) of progression from original emulsion to self-lubricating flow. (a) Original emulsion (b) Boundary B (shown in Figure 4.29) observed between emulsion layers (Observed in all cases). (c) Self lubricating flow- Emulsion layers separated by a lubricating water layer. Emulsion

Cup Convex shape

Concave meniscus ----+------Spindle Water layer

Figure 4.29 Schematic diagram of the flow structure of N-Brightstock oil emulsion during self-lubricating flow. B is the boundary of the emulsion surrounding the spindle.

Phase separation of oil and water was sometimes observed at the end of the experiments when the viscometer was disassembled. Huge droplets of water of the order of 5 to 10 J!L were also observed in the emulsion at the end of the experiment.

It should be pointed out that the convex shaped emulsion around the spindle and the concave meniscus illustrated in Figure 4.29 were observed during some of the non­ lubricating flow experiments.

104 But observations made suggest that the coalescence of the droplets in the region bordering the convex shaped emulsion layer surrounding the spindle is what determines self-lubricating flow. In other words, the convex shaped emulsion layer surrounding the spindle and the concave meniscus (Figure 4.29) are a characteristic of self-lubricating flow and sometimes non-lubricating flow but the coalescence of water droplets distinguishes the two flows.

An experiment was performed which involved shearing pure N-Brightstock oil at a spindle speed of 90.5 rpm and a temperature of 30°C. Droplets of water were gradually added to the oil with the aid of a syringe. The surface of the oil was initially level with the top of the spindle. As more drops were added, the convex shape around the spindle and the concave meniscus started to become more prominent. The results from this test seem to show that the water in the emulsion may be responsible for the formation of these two features.

The flow structures previously discussed for N-Brightstock oil were also observed in the Shellflex 810 and bitumen emulsion experiments.

It is advantageous to analyse the ratio of droplet size to gap width (or pipe diameter) as has been done previously in studies of particle migration. As previously noted in Chapter 3, researchers on single drop migration have used radius of droplet I radius of pipe ratios of about 0.02 to 0.3 (Brenner, 1966) in their Poiseuille flow migration experiments. Almost no information is available regarding water droplet sizes in bitumen froth produced at Aurora. However, Munoz and Chu (1994) did measure water droplet sizes in other sources of bitumen froth. They found water droplets of the order of 13 to 500 J.lm for hydrotransport based extraction processes. The drop diameter: pipe diameter ratio based on the 0.91 m diameter Aurora pipeline is between 0.0005· and 0.000014. This ratio is smaller than the ratios stated above by Brenner (1966).

An equivalent ratio may be calculated for Couette flow. The equivalent ratio will thus be the ratio of the radius of the droplet I radius of the gap width between the spindle and

105 cup. Equivalent ratios of the order of 0.05 are computed for Couette flow experiments of Karnis and Mason (1967); and Chan and Leal (1981). These ratios appear to be similar to the ratios of 0.02 to 0.03 used by previous researchers in Poiseuille flow experiments.

The mean water droplet diameter in the N-Brightstock oil emulsion before shearing occurs has been estimated to be of the order of 2 J.lm. The gap widths corresponding to the Z31 (15.725 mm radius), Z38 (19.01 mm radius) and Z41 (20.71 mm radius) spindles used in this research are 5.942 mm, 2.657 mm and 0.957 mm respectively. The corresponding ratios are 0.00034, 0.00075 and 0.002.

The ratios obtained for the Z31 and Z38 spindles appear to be similar to the ratios obtained above for bitumen froth in the pipeline. This similarity might be the basis for a comparison of the migration mechanism in the bitumen froth pipeline and the Couette cell experiments in this study.

The mechanisms for single droplet migration has been discussed in Chapter 3. It was stated that the mechanism for single droplet migration is probably based on the deformation of the drop, the inertia of the suspending fluid and the effect of a bounding wall. It is not known if this mechanism can be extended to experiments involving a reasonable number of droplets as was the case in this research.

Leighton and Acrivos ( 1987) state that the effect of inertia may be neglected in their Couette flow experiments of concentrated suspensions of spheres because of their small 4 particle Reynolds number of 10 • But Karnis et al. (1963) observed (single) liquid droplet migration at similar Reynold numbers although the liquid droplet was more viscous than the suspending medium.

The best approach to formulating a droplet migration mechanism is probably that of Phillips et al. (1991). But it is not known if their mechanism based on the migration of concentrated suspensions of rigid spheres is directly applicable to migration of droplets.

106 It appears that the deformation of the droplets will have to be considered in any potential mechanism.

4.11 Summary

The results from this study are summarized below:

• The variation of density with temperature for N-Brightstock oil and Shellflex 810 oil was found to be described by the following equations respectively:

p = -0.0095T2 + 0.2281T + 920.87 (4.1)

p = -0.5616T +903.77 (4.2)

where p is density in kg/m3 and T is temperature in °C

• The variation of viscosity with temperature for N-Brightstock oil and Shellflex 810 oil was found to be described by the following equations respectively:

9211.2 }l = 4.16 X 10-13 e-r- (4.7)

6896.6 Jl = 1.354 X 10-10 e_r_ (4.8)

where f.1 is the viscosity in Pa.s and T is the absolute temperature in K.

• The effects of viscous heating must be considered when viscous oils are rheologically characterised. Viscous heating effects have been shown to be proportional to the square of the angular velocity of the spindle and the effect is more pronounced with increasing spindle diameter.

• A clearly defined transition region exists between the non-lubricating flow regime and the self-lubricating flow regime. Self-lubricating flow of emulsions

107 commences at a specific shear rate and is dependent on the emulsion . temperature and water concentration.

• The critical viscometer spindle speed required to achieve self-lubricating flow of an emulsion increases with increasing temperature at constant water concentration.

• The critical viscometer spindle speed required to achieve self-lubricating flow of an emulsion decreases with increasing water concentration at a fixed emulsion temperature. An optimum water concentration of 30 wt% was discovered for N­ Brightstock emulsion. The critical spindle speed did not change at concentrations greater than 30 wt%.

• For a fixed emulsion water concentration of 30 wt%, the Initial Torque Reduction Time was found to decrease with increasing spindle speed at lower temperatures. At higher temperatures, the Initial Torque Reduction Time did not vary significantly with spindle speed.

• Self-lubricating flow was achieved with the small and medium diameter spindles but not with the large diameter spindle within the experimental limits of this study. It is possible that a minimum shear rate variation is required to achieve self-lubricating flow.

• The viscosity of the oil phase was found to be an important determinant in self­ lubricating flow of oil emulsions. It has been shown that the critical viscometer spindle speeds required to achieve self-lubricating flow of the oil emulsions are comparable when the viscosities of oils are similar for the three oils colllsidered in this study.

• Visual inspection of the exposed surface of the emulsion during self-lubricating flow revealed the presence of a convex emulsion surface adjacent to the: spindle

108 and a concave emulsion surface from the boundary of the convex surface to the cup. A thin water layer was observed in the region around 0.3 to 0.6 of the gap width between the spindle and the cup during self-lubricating flow.

109 Chapter Five

MODEL DEVELOPMENT FOR SELF-LUBRICATING FLOW IN A COUETTE CELL

Models were developed to predict the thickness of the lubricating water layer formed during self-lubricating flow. The visual observations of the N-Brightstock emulsion at the upper emulsion surface during self-lubricating flow described in Section 4.10 in Chapter 4 form the basis of these models. The first model, Model I assumes that the effective shear occurs in the water layer only while the second model, Model II assumes shear in the three distinct layers throughout the Couette cell gap. The measured torques from the experiments are used as an input variable in these models. A detailed derivation of Model II is provided in Appendix D.

A diagram representing Model I is illustrated in Figure 5.1. The three regions which are observed during self-lubricating flow are shown in Figure 5.1. The thickness of the water layer B has been exaggerated in the diagram. It separates the two emulsion layers A and C.

R1 is the radius of the spindle, R2 and R3 are the radii of the emulsion-water interface and ~ is the radius of the cup. The spindle rotates at an angular velocity ro.

110 5.1 Model I - Shearing in Water Layer Only

Cup

0) I ~ + Spindle

Figure 5.1 Schematic illustration of Model I (Shearing in water layer only). Emulsion layer- A, Water layer- B, Emulsion layer-C.

111 Model I assumes that effective shear occurs only in the water phase. There is constant shear in Emulsion layer A and no shear in Emulsion layer C. The velocity profile is illustrated in Figure 5.1.

The fluids will be assumed to be Newtonian, incompressible, at steady state, and under laminar flow conditions. The viscosity of the emulsion phases will be assumed to be the same as that of the pure oil (N-Brightstock).

Radius Rt and~ are specified. The radius of the emulsion layer A-water interface, R2, is assumed to be 0.5 of the gap width between the cup and spindle based on visual observations previously noted. It should also be noted that the model permits this position to be varied.

Radius R3 is to be determined by the model. The thickness of the water layer is thus the difference between radii R3 and R2.

Based on Figure 5.1, the boundary conditions are: At r = R., ve= Rtm At r = R2, ve = R2m At r = R3, va= 0

Atr=~, ve = 0

The solution for the velocity profile in the water phase is similar to the previous solution (Equation 2.1) given in Chapter 2 for single phase Couette flow. Since the effective shear occurs in the water phase, the torque exerted at interface R2 is thus given by:

112 (5.1)

where J.lB is the viscosity of the water phase.

113 5.2 Model II - Shearing Throughout Couette Cell Gap

Model II attempts to represent the flow through the use of a more sophisticated approach compared to Model I and is illustrated in Figure 5.2.

Cup

~

+ Spindle

Figure 5.2 Schematic illustration of Model II (Shearing in water and emulsion layers). Emulsion layers - A and C; Water layer- B.

114 Three regions A, B and Care shown in Figure 5.2. A and C represent the emulsion layers while B represents the water layer. All the assumptions for Model I are valid except that shearing is allowed to occur in all three phases.

The velocity profile for single phase Couette flow has been derived in Appendix E (Equation E.1 0). The velocity profiles for the fluids illustrated in Figure 5.2 are similar and are given by the following equations:

Fluid A, (Rt< r

(5.2)

(5. 3)

(5. 4)

where C~, C2, D~, D2, Et and E2 are constants to be determined.

Boundary conditions

At r = R2, vaA =vas; 'trllA = 'trlls

At r = R3, vas = vac; 'trlls = 'trllc

At r = Itt, voc=O

115 Detailed derivations are provided in Appendix D. The predicted velocity profiles for Model II are determined to be

(5.5)

(5.6)

The torque exerted on the spindle is

2 T = -4trp~pBR 1 RiR;R;Lm (5.7) K where

The predicted velocity profiles are illustrated in Figure 5.3. A spindle speed of90.5 rpm and a temperature of 30°C were used in the computations. This is an example of an experimental condition that resulted in self-lubricating flow in this study.

116 1.2

-+-R1

~ •• "(j 0 "i 0.6 > II WATERLAYERB tn tn cCD 0 0.4 .II "iii c CD E II c 0.2

EMULSION LAYER C

IlL. A ... A . ... - A ... 0 - - ... 0 0.2 0.4 0.6 0.8 -1 1.2 Position within gap, (r • Rt) I (R. -R1)

Figure 5.3 Predicted velocity profile for self-lubricating flow of an emulsion in a Couette cell (proceeding from spindle towards cup) from Model II. Rt = 15.2 mm

(MVIII spindle), R2 = 18.1 mm, R4= 21 mm, L = 60 mm. R3 predicted from Model II as 18.105 mm. T = 30°C, n = 90.5 rpm.

117 5.3 Comparison of Experimental Results With Model Predictions

Two typical self-lubricating flow experimental conditions (n = 90.5 rpm, T = 30°C) and (n = 362 rpm, T = 50°C) were chosen to test the models. Viscosity values from the literature will be used for water (Bennett and Myers, 1982). Radius R2 will be assumed at the midpoint between the spindle and the cup radii. Radius R3 will be determined from the model. The averaged experimental torques are computed from the torques of the several experiments conducted at these conditions. These values are found given in

Table 4.6. The thickness of the water layer is then R3-R2• The results are summarized in Table 5.1

Table 5.1 Comparison of water layer thicknesses using models I and II

Computed Water Layer Average of Thickness (mm) Temperature Spindle speed experimental \C) (rpm) torque (Nm) Modell Model II

30 90.5 0.0029 0.0058 0.0050

50 362 0.0040 0.012 0.0084

It can be observed that the thicknesses predicted by Models I and II are similar for both cases. There is a 15% difference between the water layer thicknesses computed from the models using the first experimental condition (30°C, 90.5 rpm). There is a 39% difference between the water layer thicknesses computed from the models using the second experimental condition (50°C, 362rpm).

It is difficult to estimate the exact thickness of the water layer since an accurate method of measurement could not be identified.

118 The predicted thicknesses of the water layer for a concentric cylinder flow apparatus, 0.0058 mm and 0.0116 mm are much smaller than the predictions of 0.3 to 0.4 mm for flow of bitumen froth in the 25 mm pipe by made Joseph et al. (1999) and the 2 mm thickness estimated by Neiman (1986) for flow in the 50 mm pipe.

It is also evident from the computed values for the water layer thickness that the water layer has to be very thin in order to result in the observed torque.

119 Chapter Six

CONCLUSIONS

The following conclusions can be made from this study:

• Bitumen emulsions and model oil emulsions ofN-Brightstock and Shellflex 810 exhibit similar self-lubricating flow characteristics in a Couette cell.

• A method has been developed to generate a reproducible, stable water-in-oil emulsion with viscous oils.

• Self-lubricating flow has been obtained in a concentric cylinder viscometer. Significant energy loss rates compared to that anticipated with the viscous oil alone were noted. This reduction in energy loss is similar to the effect noted with pipeline flow of heavy oils and bitumen (Kruka, 1977; Joseph et al., 1999).

• Reproducibility of the self-lubricating flow experiments was established.

• Consistent flow regime maps have been developed for the viscous oils considered in this study. The important conclusions from these flow maps include:

a) For a given temperature, water fraction and type of oil, a single critical spindle rotational speed exists. Non-lubricating flow persists below this speed and self-lubricating flow persists above this speed.

120 b) For all oils considered in this study, at a fixed water concentration, the critical spindle speeds increase with increasing temperature. This indicates that the critical speed is a function of the viscosities of the oil and water constituents. c) The critical spindle speed decreases with water concentration when other variables are held constant.

In addition, subsequent tests showed: a) When the viscosities of the two oils were adjusted using temperature to have similar viscosity, similar critical spindle speeds were measured. Therefore, for the three oils considered in this study, the type of oil does not appear to be an important determinant of critical spindle speed. b) There appears to be a minimum gap size between the concentric cylinders where self-lubricating flow will not occur. If the other variables were held constant, similar critical spindle speeds were observed for the two smaller gap sizes considered in this study.

• A set of experiments were performed to study the period in which the torque reduction takes place (initial torque reduction time period). The time period increases significantly at lower emulsion temperatures (and subsequently lower critical velocities).

• Although the structure of the self-lubricating flow structure could not be observed within the concentric cylinder apparatus, observations could be made at the air-emulsion interface. With non-lubricating flow which corresponded to a high spindle torque reading, large droplets were observed to follow a single orbital path. When the spindle speed was increased, a continuous water layer was observed which coincided with a substantial torque reduction.

121 • Although the thickness of the proposed annular layer of water could not be directly measured, it was estimated to be of the order of O.Olmm based on a theoretical model proposed in this study.

122 Chapter Seven

RECOMMENDATIONS

The following suggestions are made for possible advancement of the work conducted in this research.

• The density of the oil and water phases should be matched and experiments conducted to determine the effect of neutral buoyancy on self-lubricating flow. This might be done using the method of McKibben et al. (2000a) in which methanol was added to reduce the density of water.

• Viscometer cup and spindles with larger radial dimensions should be used to further study the influence of radial variation in shear rates on self-lubricating flow.

• Techniques should be developed to permit the flow structure to be observed.

• Emulsions with a narrow droplet size distribution but a wider range of mean sizes should be employed to determine if there is a critical droplet size required for self-lubricating flow.

• Further studies should be conducted with the model oil emulsions using pipelines to ascertain if results similar to those obtained using the Couette cell will be obtained.

123 REFERENCES

Arney, M. S., Bai, R., Guevara, E., Joseph, D. D. and Liu, K.; Friction Factor and Holdup Studies for Lubricated Pipelining- I Experiments and Correlations, Int. J. Multiphase Flow, v19, n180, pp. 1061 -1076 (1993).

Arney, M. S., Ribeiro, G. S., Guevara, E., Bai, R. and Joseph, D. D., Cement-Lined Pipes for Water Lubricated Transport of Heavy Oil, Int. J. Multiphase Flow, v22, n2, pp. 207 -221 (1996).

Bannwart, A.; Modeling Aspects of Oil-Water Core-Annular Flows, Journal of Petroleum Science and Engineering, v32, pp. 127-143 (2001).

Benayoune, M., Khezzar, L. and Al-Rumhy, M.; Viscosity of Water in Oil Emulsions, Petroleum Science and Technology, v16 (7&8}, pp. 767-784 (1998).

Bennett, C. 0. and Myers, J. E.; Momentum, Heat and Mass Transfer, McGraw- Hill Inc., U.S.A., 3rd ed., 1982.

Bhardwaj, J. and Hartland, S.; Kinetics of Coalescence of Water Droplets in Water-in­ Crude Oil Emulsions, Journal of Dispersions Science and Technology, v15, n2, pp. 133 - 146 (1994).

Bird, R. B., Stewart, W. E. and Lightfoot, E. N.; Transport Phenomena, John Wiley and Sons, Inc., New York, 1960.

Brenner, H., Hydrodynamic Resistance of Particles at Small Reynolds Numbers, Advances in Chemical Engineering, v6, pp. 377-403 (1966).

Chan, P. C.-H. and Leal, L. G.; An Experimental Study of Drop Migration in Shear Flow between Concentric Cylinders, Int. J. Multiphase Flow, v7, pp. 83-99 (1981).

Charles, M. E.; The Reduction of Pressure Gradients in Oil Pipelines: Experimental Results for The Stratified Flow ofA Heavy Crude Oil and Water, 11th Annual Technical Meeting, Petroleum and Natural Gas Division, C.I.M., and the 6th Annual Meeting, Rocky Mountains Sections, Society of Petroleum Engineers, A.I.M.E., Calgary, Transactions Volume LXIII, pp. 306 - 310, May 1960.

Charles, M. E., Govier, G. W. and Hodgson, G. W.; The Horizontal Pipeline Flow of Equal Density Oil-Water Mixtures, Can. J. Chern. Eng., pp. 27-36, February, 1961.

Chilingarian, G.V. and Yen, T. F.; Bitumens, Asphalts and Tar Sands, Elsevier Scientific Publishing Company, 1978.

124 Dealy, J. M.; Rheological Properties of Oil Sand Bitumens; Can. J. Chern. Eng.; v57, pp. 677-683, December 1979.

Furuta, T., Tsujimoto, S., Toshima, M., Okazaki, M. and Ryozo,T.; Concentration Distribution of Particles in Solid-Liquid Two-Phase Flow Through Vertical Pipe; Memoirs of the Faculty of Engineering, Kyoto University, Kyoto, Japan, v39, n4, pp.605- 615, (1977).

Gauthier, F., Goldsmith, H. L. and Mason, S. G.; Particle Motions in Non-Newtonian Media I: Couette Flow, Rheol. Acta., v10, pp. 344-364 (1971).

Goldsmith, H. L. and Mason, S. G.; The Flow of Suspensions Through Tubes. Part 1. Single Spheres, Rods and Discs; Journal of Colloid Science, v17, pp. 448-476 (1962).

Government of Alberta, Introduction to Oil Sands, at URL htto://www .energy. gov.ab.ca/com/Sands/lntroduction/Oil Sands.htm, (2003 ).

Hashimoto, C., Cristobal, G., Nicolas, A., Panizza, P., Rouch, J. and Ushiki, H.; Low­ Cost Viscometer Based on Energy Dissipation in Viscous Liquids, Meas. Sci. Technol.; v12, pp. 1 - 5 (2001 ).

Ho, B. P. and Leal, L. G.; Inertial Migration of Rigid Spheres in Two-Dimensional Unidirectional Flows, J. Fluid Mech., v 65, part 2, pp. 365-400 (1974)

Isaacs, J.D. and Speed, J. B.; Method ofPiping Fluids; United States Patent, Patent No. 759,374 (1904).

Joseph, D. D., Bai, R., Chen, K. P. and Renardy, Y. Y.; Core-Annular Flows, Annu. Rev. Fluid Mech., v29, pp. 65-90 (1997).

Joseph, D. D.; Lubricated Transport of Viscous Materials, Fluid Mechanics and its Applications, v43, pp. 1 -24 (1998).

Joseph, D. D., Bai, R., Mata, C., Sury, K. and Grant, C.; Self-Lubricated Transport of Bitumen Froth, J. Fluid Mech., v386, pp. 127-148 (1999).

Karnis, A., Goldsmith, H. and Mason, S. G.; Axial Migration of Particles in Poiseuille Flow; Nature, v200, pp. 159-160 (1963).

Karnis, A. and Mason, S.G.; Particle Motions in Sheared Suspensions XXIII Wall Migration ofFluid Drops, J. Colloid Interface Sci., v24, pp. 164-169 (1967).

Khan, M. R., Rheological Properties of Heavy Oils and Heavy Oil Emulsions, Energy sources, v18, pp. 385-391 (1996).

125 Kruka, V. R., Method for Establishing Core-Flow in Water-in-Oil Emulsions or Dispersions, United States Patent, Patent No. 4,047,539, September 13, 1977.

Leal-Calderon, F., Gerhardi, B., Espert, A., Brossard, F., Alard, V., Tranchant, J.F., Stora, T. and Bibette, J., Aggregation Phenomena in Water-in-Oil Emulsions, Langmuir, v12, n4, pp. 872- 874, Feb 21, 1996.

Leighton, D. and Acrivos, A., The Shear-Induced Migration of Particles in Concentrated Suspensions, J. Fluid Mech., v181, pp. 415-439 (1987).

Litzenberger, C., Private Communication, 2002.

Mankowski, P., S. Ng, R. Siy, J. Spence and P. Stapleton, Syncrude's Low Energy Extraction Process: Commercial Implementation, Proc. 31st Ann. Can. Min. Processors, Ottawa, ON, Canada, Jan. 19-21, 1999, Can. Inst. ofMining, Metallurgy and Petroleum, Ottawa, ON, pp. 153-181, (1999).

McKibben, M. J ., Gillies, R. G. and Shook, C. A.; A Laboratory Investigation of Horizontal Well Heavy Oil-Water Flows; Can J. Chern. Eng.; v78, pp. 743 - 751 (2000a).

McKibben, M. J., Gillies, R. G. and Shook, C. A.; Predicting Pressure Gradients in Heavy Oil-Water Pipelines, Can. J. Chem.Eng., v78, pp. 752-756 (2000b).

Munoz, V. and Chu, A.; Preliminary Confocal Laser Scanning Microscopy of Bitumen Froth, Division Report WRC 94-54 (CF), Canada Centre for Mineral and Energy Technology, 1994.

Neiman, 0.; Froth Pipelining Tests, Syncrude Canada Ltd. Res. Dev. Prog. Report 15(1), pp. 373-407, 1986.

Neiman, 0., Sury, K., Joseph, D. D., Bai, R. and Grant, C.; Process for Pumping Bitumen Froth through A Pipeline, United States Patent, Patent No. 5,988, 198, Nov. 23,1999.

Nunez, G., Briceno, M., Mata, C., Rivas, H. and Joseph, D. D.; Flow Characteristics of Concentrated Emulsions of Very Viscous Oil in Water, J. Rheol., v.40, n3, pp. 405 - 423, May/June 1996.

Nunez, G. A., Rivas, H. J., and Joseph, D. D.; Drive to Produce Heavy Crude Prompts Variety of Transportation Methods, Oil Gas J., pp. 59-68, Oct. 26, 1998.

Oliemans, R. V. A. and Ooms, G.; Core-Annular flow of Oil and Water through A Pipeline, Multiphase Sci. Technol.; v2, pp.427-76 (1986).

126 Ooms, G., Segal, A., Van der Wees, A. J., Meerhoff, R. and Oliemans, R. V. A.; A Theoretical Model for Core-Annular Flow of a Very Viscous Oil Core and a Water Annulus through a Horizontal Pipe, Int. J. Multiphase flow, v10, n1, pp. 41 -60 (1984).

Otsubo, Y. and Prud'homme, R. K.; Effect of Drop Size Distribution on the Flow Behaviour of Oil-in-Water Emulsions, Dept. of Image Science, Chiba University, Chiba, Japan, Rheologica Acta, v33, n4, pp. 303-306 (1994).

Pal, R. and Rhodes, E.; Viscosity/Concentration Relationships for Emulsions, Journal of Rheology, v33, n7, pp. 1021-1045, Oct 1989.

Pal, R., Yan, Y. and Masliyah, J.; Rheology of Clay-in-Oil Suspensions with Added Water Droplets, Chemical Engineering Science, v47, n5, pp. 967-970, (1992).

Pal, R.; Shear Viscosity Behaviour of Emulsions of Two Immiscible Fluids, Journal of Colloid and Interface Science, v.225, pp. 359-366 (2000).

Phillips, R. J ., Armstrong, R. C. and Brown, R. A.; A Constitutive Equation for Concentrated Suspensions that Accounts for Shear-Induced Particle Migration, Phys. Fluids A, v4, n1, pp. 30-40, January 1992.

Russell, T. W. F., Hodgson, G. W. and Govier, G. W.; Horizontal Pipeline Flow of Mixtures ofOil and Water; Can. J. Chern. Eng., v37, pp. 9-17 February, 1959.

Sanchez, L. and Zakin, J.; Transport of Viscous Crudes as Concentrated Oil-in-Water Emulsions, Ind. Eng. Chern. Res., v33, pp. 3256-3261 (1994).

Sanders, R. S., Halferdahl, G. and Shook, C. A.; Correlation of Slurry Sedimentation, Transportation and Deposition Characteristics for Oil Sand Tailings, Proc. 11th Int. Conf. on Transport & Sedimentation of Solid Particles, 9-12 Sept., Ghent, Belgium, 2002.

Sanders, R.S.; Private Communication, 2003.

Sanders, R. S., Ko, T., Bai, R. and Joseph, D. D.; Factors Governing Friction Losses in Self-Lubricated Transport of Bitumen Froth: 1. Water Release, Submitted to Can J. Chern. Eng.; To be published April2004.

Schaan, J., Sanders, R. S., Litzenberger, C., Gillies, R. G. and Shook, C. A.; Measurements of Heat Transfer Coefficients in Pipeline Flow of Athabasca Bitumen Froth; Presented at 3rd North American Conference on Multiphase Technology, Banff, Canada, 6 - 7 BHR Group Limited, pp. 25-39, June 2002.

127 Schramm, L. L. and Kwak, J. C. T.; The Rheological Properties of an Athabasca Bitumen and Some Bituminous Mixtures and Dispersions; Journal of Canadian Petroleum Technology; v27, n1, pp.26- 35, Jan-Feb 1988.

Segre, G. and Silberberg, A.; Behaviour of Macroscopic Rigid Spheres in Poiseuille Flow - Part 1. Determination ofLocal Concentration by Statistical Analysis ofParticle Passages Through Crossed Light Beams; J. Fluid Mech.; v14, pp. 115-135, 1962a.

Segre, G. and Silberberg, A.; Behaviour of Macroscopic Rigid Spheres in Poiseuille Flow- Part 2. Experimental Results and Interpretation; J. Fluid Mech.; v14, pp. 136- 157, 1962b.

Shaw, R. C., L. L. Schramm and J. Czarnecki, Suspensions in the Hot Water Flotation Process for Canadian Oil Sands, in Suspensions: Fundamentals and Applications in the Petroleum Industry, L. L. Schramm (Ed.), ACS Advances in Chemistry, Vol. 251, American Chemical Society: Washington, DC, pp. 639-675, (1996).

Shennan, P., Rheological Properties of Emulsions, Encyclopedia of Emulsion Technology, P.Becher (ed.), vi, Marcel Dekker Inc., New York, pp. 405-439, 1988.

Shook, C. A. and Roco, M.C.; Slurry Flow : Principles and Practice; Butterworth­ Heineman, USA, p. 58, 1991.

Shook, C. A., Gillies, R. G. and Sanders, R. S.; Pipeline Hydrotransport: With Applications in the Oil Sand Industry, SRC Publications, No. 11508-1E02, Saskatoon, SK, 2002.

Smart, J. R. and Leighton, D. T., Measurement of the Drift of a Droplet Due to The Presence ofA Plane, Phys. Fluids A 3(1), pp. 21-28, January 1991.

Sumner et al., Froth Report (2003), to be published.

Syncrude Canada, at URL http://www.syncrude.com/who we are/01 04 l.html http://www.svncrude.com/who we are/01 04 2.html, (2003)

Vander Kloet, J., Schramm, L. L. and Shelfantook, B.; The Influence of Bituminous Froth Components on Water - in - Oil Emulsion Stability as Determined by The Micropipette Technique, Colloids and Surfaces A: Physicochemical and Engineering Aspects, vl92, pp. 15-24 (2001).

Western Oil Sands, Oil Sands Introduction, at URL http://westernoilsands.com/html/business/introduction.html, (2003 ).

128 World Energy Council, Natural Bitumen and Extra-Heavy Oil, at URL http://www. worldenergy.org/wec-geis/publications/reports/ser/bitumen/bitumen.asp (2003).

Yaghi, B., Benayoune, M. and Al-Bemani, A.; Viscosity of Water-Oil Emulsions with Added Nanoparticles, Pet. Sci. Techno!., v19 (3&4), pp. 373-386, (2001).

Yan, Y., Pal, R., and Masliyah, J.; Rheology of Oil-in-Water Emulsions with Added Solids, Chemical Engineering Science, v46, n4, pp. 985-994, (1991).

Yan, N. and Masliyah, J.; Characterization and Demulsification ofSolids-Stabilized Oil in Water Emulsions Part 1. Partitioning of Clay Particles and Preparation of Emulsions, Colloids and Surfaces A: Physicochemical and Engineering Aspects, v96, n3, pp. 229-242, Mar 1995.

129 APPENDIX A Oil Density and Viscosity Data

Oil density data is presented in Tables A.1 and A.2. Oil viscosity data is presented in Tables A.3 and A.4. Newtonian and Bingham model fits to experimental viscosity data and linear plots of the Arrhenius equation for viscosity are presented in Figures A.1 to A.6

Table A.l Measured N - Brightstock oil density as a function of temperature

Temperature Density 3 \C) (kglm )

10 922.02 20 921.82 25 920.84 30 919.47 35 916.72 40 914.75 45 911.61 50 908.86

130 Table A.2 Measured Shellflex 810 oil density as a function of temperature Temperature Density \C) (kglm3)

10 898.42 15 895.03 20 892.29 25 889.59 30 886.73 35 883.92 40 880.95 45 878.22 50 875.54

Table A.3 Measured N-Brightstock oil viscosity as a function of temperature (J.lp is plastic viscosity and 'ty is yield stress).

Calculated Viscosity

Newtonian Binpam

Temperature \C) JJ. (Pa· s) p.., (Pa· s) 'l:v (Pa) 10 64.97 62.03 6.98 15 32.50 31.68 1.97 22.5 14.72 14.36 0.86 30 6.54 6.37 0.39 40 2.42 2.38 0.11

50 1.04 1.04 0.01

131 Table A.4 Measured Shellflex 810 oil viscosity as a function of temperature (J.tp is plastic viscosity and 'ty is yield stress).

Calculated Viscosity

Newtonian Bingham

Temperature tC) J.L (Pa· s) 1Jo (Pa· s) ~v(Pa)

0 13.63 13.56 0.17

15 3.15 3.14 0.02

30 0.98 0.27 0.01

50 0.27 6.37 0.39

132 10.------~

9

8

--! 7 "C l! -e ·c:;~ 0 5 ~ ; 4 • 1\Aeasured torque data :; C) - Unear Newtonian model fit ~ 3

2

1

0+-----~------~----~------~----~------~----~------~ 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 Torque/Length (N-mlm)

Figure A.l Newtonian model fit for measured torque and angular velocity data of Cannon 8600 viscosity standard obtained using Z31 Ti spindle (15.725 mm radius) at a temperature of 20°C (Correlation coefficient of linear Newtonian model fit, R2 = 0.9999). Spindle speed range = 2 to 10 rpm.

133 10

9

8 .-.. .!!!.? "'0 ....('G ;:e..... ·u .25 G) > _!4 • l\l1easured torque data :::s C) - Unear Bing&n model fit C3

1

0 0.00 0.02 0.04 0.00 0.08 0.10 0.12 0.14 0.16 Torque/Length (N-m'm)

Figure A.2 Bingham model fit for measured torque and angular velocity data of Cannon 8600 viscosity standard obtained using Z31 Ti spindle (15.725 mm radius) at a 2 temperature of 20°C (Correlation coefficient of linear Newtonian model fit, R = 0.9999). Spindle speed range= 2 to 10 rpm.

134 3.------~

2.5~------~L---~

-tA cO ~ 1.5+------~~------~ :::1 s::::: • Measured viscosity data

-Linear fit

0+-----~~------~------~------~------~------~------~ 0.00305 0.0031 0.00315 0.0032 0.00325 0.0033 0.00335 0.0034 1 1/T (K- )

Figure A.3 Natural logarithm of the measured Newtonian viscosities of N - Brightstock oil as a function of the inverse of the absolute Temperature (based on 2 Equation 4.5). Correlation coefficient of linear fit, R = 0.9992.

135 3

2.5 ~

2 /

1.5 /

1 / t/) -ni e:.. 0.5 / ::1 .5 0 / • Measured viscosity data - -Linear fit -0.5 /

-1 /

-1.5

-2 I 0.003 0.0031 0.0032 0.0033 0.0034 0.0035 0.0036 0.0037 1/T (1<"1)

Figure A.4 Natural logarithm of the measured Newtonian viscosities of Shell flex 810 oil as a function of the inverse of the absolute Temperature (based on Equation 4.5). 2 Correlation coefficient of linear fit, R = 0.9977.

136 1.2.------~

1.0

.!!- ~0.8 ._.,.... ·ub .20.6 ~ .... • Measu'ed torque data ftl :i - Unear Nevttonian model fit CA0.4 c

0.2

0.0+---~~--~----~----~--~~--~----~----~----~--~ 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 0.0016 0.0018 0.0020 Torque/Length (N-mlm)

Figure A.S Newtonian model fit for measured torque and angular velocity data of Shellflex 810 oil obtained using Z31 Ti spindle (15.725 mm radius) at a temperature of 50°C (Correlation coefficient of linear Newtonian model fit, R2 = 1). Spindle speed range = 2 to 10 rpm.

137 1.2.------~

1.0 ...... ! "0 ..__.f!0.8 b ·c:; .20.6 Q) >... ..!! 61>.4 c <(

0.2

0.0+---~----~----~--~----~----~--~----~----~--~ 0.00 0.02 0.04 0.00 0.~ 0.10 0.12 0.14 0.16 0.18 0.20 T~{Nm'm)

Figure A.6 Newtonian model fit for measured torque and angular velocity data of coker-feed Bitumen obtained using Z31 Ti spindle (15.725 mm radius) at a temperature of 50°C (Correlation coefficient of linear Newtonian model fit, R2 = 0.9999). Spindle

speed range = 2 to 10 rpm.

138 APPENDIX B Couette Flow Experimental Data Details of all the Couette flow experimental runs are presented in Table B.l. Reproducibility of the Couette flow experimental results obtained with 30 wt% water­ in-N-Brightstock emulsion is presented in Table B.2. Couette flow experimental results are presented in Tables B.4 to BlO. The flow maps for N-Brightstock emulsions with varying water concentrations are presented in Figures B.l to B.4.

139 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions

tRatio of Emulsion Water Spindle Type of Initial Final initial Run Shearing Measuring concentration Temperature speed Lubricating torque torque to final No. Oil Device Spindle (wt 0/o) \C) (rpm) Flow (N-m) (N-m) torque *64a N - Brightstock HaakeRV3 MVIII 25 22.5 16 NLF 0.0125 0.0129 0.97 64b N - Brightstock Haake RV3 MVIII 25 22.5 22.6 NLF 0.0174 0.0179 0.98 64c N - Brightstock HaakeRV3 MVIII 25 22.5 32 NLF 0.0243 0.0246 0.99 64d N - Brightstock HaakeRV3 MVIII 25 22.5 45.2 NLF 0.0333 0.0331 1.01 '

64e N - Brightstock HaakeRV3 MVIII 25 22.5 64 NLF 0.0431 0.0057 7.59 I 65 N - Brightstock HaakeRV3 MVIII 25 22.5 64 MLF 0.0417 0.0096 4.36 66 N - Brightstock HaakeRV3 MVIII 25 22.5 45.2 NLF 0.0333 0.0341 0.98 67 N - Brightstock HaakeRV3 MVIII 30 22.5 64 SLF 0.0385 0.0013 30.19 I--' .,J::.. 68 N - Brightstock HaakeRV3 MVIII 30 22.5 45.2 SLF 0.0323 0.0025 13.20 0 69 N - Brightstock HaakeRV3 MVIII 35 22.5 64 SLF 0.0380 0.0012 32.29 70 N- Brightstock HaakeRV3 MVIII 35 22.5 45.2 SLF 0.0294 0.0033 8.82 71 N - Brightstock HaakeRV3 MVIII 35 22.5 32 SLF 0.0230 0.0014 16.63 72 N - Brightstock Haake RV3 MVIII 35 22.5 22.6 NLF 0.0203 0.0203 1.00 73a N - Brightstock HaakeRV3 MVIII 25 30 16 NLF 0.0074 0.0074 1.00 73b N - Briglltstock HaakeRV3 MVIII 25 30 22.6 NLF 0.0100 0.0096 1.05 - -·- -·------·--- *Run numbers e.g. 64a, 82b indicate that the Stepwise spindle speed procedure was used. Abbreviations NLF represents Non­ Lubricating Flow, MLF represent Marginal Lubricating Flow and SLF represents Self-Lubricating Flow. Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Emulsion ~atio oj Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating to1·que torque to final Run No. Oil Device Spindle (wt o/o) rc) (rpm) Flow (N-m) (N-m) torgue 73c N - Brightstock HaakeRV3 MVIII 25 30 32 NLF 0.0131 0.0127 1.03 73d N - Brightstock HaakeRV3 MVIII 25 30 45.2 NLF 0.0172 0.0167 1.03 73e N - Brightstock HaakeRV3 MVIII 25 30 64 NLF 0.0225 0.0218 1.03 73f N - Brightstock HaakeRV3 MVIII 25 30 90.5 NLF 0.0292 0.0257 1.13 74 N- Brightstock HaakeRV3 MVIII 25 30 128 SLF 0.0448 0.0037 12.19 75a N - Brightstock HaakeRV3 MVIII 30 30 16 NLF 0.0093 0.0091 1.03

...... 75b N - Brightstock HaakeRV3 MVIII 30 30 22.6 NLF 0.0123 0.0118 1.04 ~ ...... 75c N - Brightstock HaakeRV3 MVIII 30 30 32 NLF 0.0159 0.0154 1.03 75d N - Brightstock HaakeRV3 MVIII 30 30 45.2 NLF 0.0206 0.0197 1.04 75e N - Brightstock HaakeRV3 MVIII 30 30 64 NLF 0.0260 0.0245 1.06

75f N - Brightstock HaakeRV3 MVIII 30 30 90.5 SLF 0.0319 0.0034 9.29 1 76 N - Brightstock Haake RV3 MVIII 30 30 90.5 SLF 0.0389 0.0018 21.15 77 N - Brightstock HaakeRV3 MVIII 30 30 64 MLF 0.0314 0.0032 9.85 78a N - Brightstock HaakeRV3 MVIII 35 30 16 NLF 0.0086 0.0086 1.00 78b N- Brightstock HaakeRV3 MVIII 35 30 22.6 NLF 0.0113 0.0110 1.02 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt%} rc) (rpm) Flow (N-m) (N-m) torque 78c N- Brightstock HaakeRV3 MYIII 35 30 32 NLF 0.0144 0.0140 1.03 78d N- Brightstock HaakeRV3 MY III 35 30 45.2 NLF 0.0179 0.0160 1.12 78e N - Brightstock HaakeRV3 MY III 35 30 64 SLF 0.0201 0.0018 11.39 79 N - Brightstock HaakeRV3 MY III 35 30 64 MLF 0.0319 0.0091 3.51 80 N - Brightstock HaakeRV3 MYIII 35 30 90.5 SLF 0.0349 0.0019 18.38 81 N- Brightstock RS 150 Z31 Ti 35 30 64 SLF 0.0149 0.0019 7.68 ...... ~ 82a N- Brightstock HaakeRV3 MY III 30 22.5 16 NLF 0.0172 0.0171 1.01 N 82b N- Brightstock HaakeRV3 MYIII 30 22.5 22.6 NLF 0.0221 0.0215 1.03 82c N - Brightstock HaakeRV3 MY III 30 22.5 32 SLF 0.0277 0.0017 16.14 83 N- Brightstock HaakeRV3 MY III 30 22.5 32 SLF 0.0247 0.0025 10.10 84a N- Brightstock HaakeRV3 MY III 30 40 22.6 NLF 0.0048 0.0047 1.02 84b N - Brightstock HaakeRV3 MY III 30 40 32 NLF 0.0061 0.0064 0.96 84c N - Brightstock Haake RV3 MY III 30 40 45.2 NLF 0.0085 0.0086 0.99 84d N- Brightstock HaakeRV3 MY III 30 40 64 NLF 0.0110 0.0111 1.00 84e N - Brightstock HaakeRV3 MY III 30 40 90.5 NLF 0.0146 0.0147 0.99 84f N- Brightstock HaakeRV3 MY III 30 40 128 NLF 0.0191 0.0186 1.03 - Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

I

Emulsion Ratio o• Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt o/o) C0C) (rpm) Flow (N-m) (N-m) torque 84g N - Brightstock HaakeRV3 MVIII 30 40 181 SLF 0.0243 0.0039 6.19 85 N - Brightstock HaakeRV3 MY III 30 40 128 MLF 0.0181 0.0130 1.40 86 N - Brightstock HaakeRV3 MVIII 30 40 181 SLF 0.0260 0.0016 16.56 87 N - Brightstock HaakeRV3 MVIII 30 40 256 SLF 0.0254 0.0025 10.38 88 N - Brightstock HaakeRV3 MVIII 30 40 512 SLF 0.0270 0.0022 12.22 89 N - Brightstock HaakeRV3 MVIII 30 22.5 22.6 NLF 0.0201 0.0174 1.15 90 N - Brightstock HaakeRV3 MYIII 30 40 128 MLF 0.0194 0.0096 2.03 1--' 91a N - Brightstock HaakeRV3 MYIII 30 50 45.2 NLF 0.0044 0.0045 0.97 ~ w 91b N - Brightstock HaakeRV3 MVIII 30 50 64 NLF 0.0056 0.0054 1.05 91c N - Brightstock HaakeRV3 MYIII 30 50 90.5 NLF 0.0069 0.0074 0.93 91d N - Brightstock HaakeRV3 MY III 30 50 128 NLF 0.0088 0.0088 1.00

91e N - Brightstock HaakeRV3 MY III 30 50 181 NLF 0.0113 0.0110 1.02 i 91f N - Brightstock HaakeRV3 MY III 30 50 256 NLF 0.0137 0.0098 1.40 91g N - Brightstock HaakeRV3 MVIII 30 50 362 SLF 0.0105 0.0027 3.91 92 N - Brightstock HaakeRV3 MY III 30 50 90.5 NLF 0.0078 0.0105 0.74 93 N- Brightstock HaakeRV3 MY III 30 50 128 NLF 0.0105 0.0091 1.16 ------Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Emulsion ~atio of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final! Run No. Oil Device Spindle (wt 0/o) (oC) (rpm) Flow (N-m) (N-m) torque 1 94 N - Brightstock HaakeRV3 MVIII 30 50 181 NLF 0.0127 0.0118 1.08

95 N- Brightstock HaakeRV3 MVIII 30 50 256 NLF 0.0196 0.0137 1.43 ! 96 N- Brightstock HaakeRV3 MVIII 30 50 362 SLF 0.0216 0.0042 5.18 1 97 N- Brightstock HaakeRV3 MVIII 30 50 512 SLF 0.0299 0.0049 6.10 98 N - Brightstock RS 150 Z31 Ti 25 30 64 NLF 0.0218 0.0190 1.14 99 N - Brightstock RS 150 Z31 Ti 30 15 16 NLF 0.0327 0.0253 1.29 100 N - Brightstock RS 150 Z31 Ti 30 15 22.6 SLF 0.0468 0.0053 8.81 101 N - Brightstock RS 150 Z31 Ti 30 15 32 SLF 0.0468 0.0059 8.00 """"' t 102 N - Brightstock RS 150 Z31 Ti 30 30 128 SLF 0.0266 0.0022 11.87

103 N - Briglltstock RS 150 Z31 Ti 30 15 45.2 SLF 0.0444 0.0028 15.73 i 104 N - Brightstock RS 150 Z31 Ti 30 15 8 NLF 0.0215 0.0184 1.17 105 N - Brightstock RS 150 Z31 Ti 30 30 181 SLF 0.0270 0.0025 10.75 106 N - Brightstock HaakeRV3 MVIII 30 50 256 NLF 0.0178 0.0135 1.32 107 N - Brightstock HaakeRV3 MVIII 30 50 362 SLF 0.0221 0.0026 8.33 108 N - Brightstock HaakeRV3 MVIII 30 50 256 NLF 0.0186 0.0137 1.36 109--._N - Brightstock Haake RV3 MVIII 30 50 362 NLF 0.0257 0.0178 1.44 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

I Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to fmal

Run No. Oil Device Spindle (wt%) \C) (rpm) Flow (N-m) (N-m) torque 1

110 N- Brightstock HaakeRV3 MVIII 30 50 512 SLF 0.0235 0.0015 15.97 i

111 N - Brightstock RS 150 Z31 Ti 30 50 362 SLF 0.0130 0.0026 5.05 i

112 N - Brightstock HaakeRV3 MVIII 30 30 90.5 SLF 0.0299 0.0034 8.71 I 113 N - Brightstock HaakeRV3 MVIII 30 30 90.5 SLF 0.0319 0.0017 18.57 114a N - Brightstock RS 150 Z41 Ti 30 22.5 8 NLF 0.0477 0.0459 1.04 114b N - Brightstock RS 150 Z41 Ti 30 22.5 16 NLF 0.0731 0.0585 1.25

~ 114c N - Brightstock RS 150 Z41 Ti 30 22.5 22.6 NLF 0.0723 0.0586 1.23 ~ v.. 115a N - Brightstock RS 150 Z38 Ti 30 22.5 8 NLF 0.0231 0.0211 1.09 115b N - Brightstock RS 150 Z38Ti 30 22.5 16 NLF 0.0358 0.0351 1.02 liSe N - Brightstock RS 150 Z38 Ti 30 22.5 22.6 NLF 0.0450 0.0400 1.12 115d N- Brightstock RS 150 Z38Ti 30 22.5 32 SLF 0.0500 0.0141 3.54 115e N - Brightstock RS 150 Z38Ti 30 22.5 45.2 SLF 0.0081 0.0061 1.33 115f N - Brightstock RS 150 Z38 Ti 30 22.5 64 SLF 0.0062 0.0052 1.19 116 HaakeRV3 MVIII 30 30 90.5 ,45.2 0.0307 ------·· - -'-~ - Brigh_!~~9Ck -~--·--·----·- -- Table B.l Couette flow experimental results for N-Brightstock, Shell flex 810 and Bitumen emulsions ( cont' d)

Ratio Emulsion of

Water Spindle Type of Initial Final initial j

Shearing Measuring concentration Temperature speed Lubricating torque torque to final I Run No. Oil Device Spindle (wt 0/o) \C) (rpm) Flow (N-m) (N-m) torque

117 N - Brightstock HaakeRV3 MVIII 30 50 362 SLF 0.0218 0.0029 7.42 1 118 N - Brightstock RS 1SO Z38 Ti 30 22.S 64 NLF 0.0322 0.0194 1.67 119 N - Brightstock HaakeRV3 MVIII 30 30 64 MLF 0.0240 0.0032 7.S4 120a N - Brightstock RS 1SO Z38 Ti 30 40 8 NLF 0.0048 0.0045 1.08 120b N - Brightstock RS 150 Z38 Ti 30 40 16 NLF 0.0081 0.0076 1.06 120c N- Brightstock RS 1SO Z38Ti 30 40 22.6 NLF 0.0102 0.0096 1.06

"""""'~ 120d N - Brightstock RS 1SO Z38 Ti 30 40 32 NLF 0.0128 0.0120 1.07 0'\ 120e N- Brightstock RS 1SO Z38 Ti 30 40 4S.2 NLF 0.0158 0.0133 1.18 120f N - Brightstock RS 1SO Z38 Ti 30 40 64 NLF 0.017S 0.0078 2.22 120g N - Brightstock RS 1SO Z38Ti 30 40 90.S NLF 0.0098 0.0098 1.00 120h N - Brightstock RS ISO Z38 Ti 30 40 128 NLF O.OI24 0.009S 1.30 120i N - Brightstock RS ISO Z38 Ti 30 40 I8I NLF O.OliO 0.007I l.SS I20j N - Brightstock RS 1SO Z38 Ti 30 40 2S6 SLF 0.0088 0.0026 3.44 I20k N - Brightstock RS I50 Z38Ti 30 40 362 SLF 0.0024 0.0021 l.IO 121 N ~ Brightstock RS ISO Z38 Ti 30 22.5 4S.2 SLF 0.0339 0.0048 7.06 ~- Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. on Device Spindle (wt o/o) rc) (rpm) Flow (N-m) (N-m) torque 122 N - Brightstock HaakeRV3 MYIII 30 50 362 SLF 0.0201 0.0020 10.25 123a N - Brightstock RS 150 Z41 Ti 30 40 8 NLF 0.0114 0.0108 1.05 l23b N - Brightstock RS 150 Z41 Ti 30 40 16 NLF 0.0197 0.0189 1.04 123c N- Brightstock RS 150 Z41 Ti 30 40 22.6 NLF 0.0252 0.0242 1.04 123d N- Brightstock RS 150 Z41 Ti 30 40 32 NLF 0.0322 0.0305 1.06 123e N- Brightstock RS 150 Z41 Ti 30 40 45.2 NLF 0.0400 0.0369 1.08 ~ ,.J:::.. -....) 123f N - Brightstock RS 150 Z41 Ti 30 40 64 NLF 0.0479 0.0420 1.14 123g N - Brightstock RS 150 Z41 Ti 30 40 90.5 NLF 0.0555 0.0507 1.09 123h N - Brightstock RS 150 Z41 Ti 30 40 128 NLF 0.0682 0.0641 1.06 124 N - Brightstock RS 150 Z38 Ti 30 22.5 64 MLF 0.0358 0.0187 1.92 125 N - Brigh_tstock HaakeRV3 MYIII 30 30 32 NLF 0.0152 0.0137 1.11 126 N - Brightstock HaakeRV3 MY III 30 30 32, ....90.5 127 N - Brightstock HaakeRV3 MVII 30 40 64, ... 256 NLF 128a Shell flex 81 0 HaakeRV3 MYIII 30 30 64 NLF 0.0066 0.0065 1.02 128b Shell flex 810 HaakeRV3 MYIII 30 30 90.5 NLF 0.0081 0.0086 0.94 Table B.l Couette flow experimental results for N-Brightstock, Shell flex 810 and Bitumen emulsions ( cont' d)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to fmal Run No. Oil Device Spindle (wt o/o) COC) (rpm) Flow _{N-m} (N-m) torque 128c Shell flex 81 0 HaakeRV3 MY III 30 30 128 NLF 0.0117 0.0118 1.00 128d Shell flex 81 0 HaakeRV3 MY III 30 30 181 NLF 0.0159 0.0159 1.00 128e Shell flex 810 HaakeRV3 MY III 30 30 256 NLF 0.0218 0.0220 0.99 128f Shell flex 81 0 HaakeRV3 MY III 30 30 362 NLF 0.0292 0.0275 1.06 128g Shellflex 810 HaakeRV3 MY III 30 30 512 NLF 0.0363 0.0042 8.71 ~ Shellflex 810 RS 150 Z31 Ti 30 0 8 NLF 0.0108 0.0103 1.05 -00 129a 129b Shell flex 810 RS 150 Z31 Ti 30 0 16 NLF 0.0192 0.0184 1.04 129c Shellflex 810 RS 150 Z31 Ti 30 0 22.6 NLF 0.0252 0.0245 1.03 129d Shellflex 810 RS 150 Z31 Ti 30 0 32 NLF 0.0336 0.0324 1.04 129e Shell flex 81 0 RS 150 Z31 Ti 30 0 45.2 NLF 0.0440 0.0422 1.04 129f Shell flex 810 RS 150 Z31 Ti 30 0 64 SLF 0.0570 0.0143 3.99 129g Shell flex 810 RS 150 Z31 Ti 30 0 90.5 SLF 0.0033 0.0032 1.03 129h Shell flex 810 RS 150 Z31 Ti 30 0 128 SLF 0.0052 0.0071 0.73 130a Shell flex 810 HaakeRY3. MVIII 30 50 90.5 NLF 0.0025 0.0025 1.00 ------Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt o/o) \C) (rpm) Flow (N-m) (N-m) torgue 130b Shell flex 81 0 HaakeRV3 MYIII 30 50 128 NLF 0.0033 0.0033 1.01 130c Shell flex 810 HaakeRV3 MYIII 30 50 181 NLF 0.0047 0.0054 0.86 130d Shell flex 810 HaakeRV3 MYIII 30 50 256 NLF 0.0065 0.0065 1.01 130e Shellflex 810 HaakeRV3 MVIII 30 50 362 NLF 0.0090 0.0089 1.02 130f Shell flex 810 HaakeRV3 MYIII 30 50 512 NLF 0.0122 0.0121 1.01 ~ -\0 130g Shell flex 810 HaakeRV3 MVIII 30 50 724 NLF 0.0126 0.0164 0.77 131a N- Brightstock HaakeRV3 MVIII 10 22.5 16 NLF 0.0105 0.0101 1.04 131b N - Brightstock HaakeRV3 MY III 10 22.5 22.6 NLF 0.0136 0.0132 1.03 131c N - Brightstock HaakeRV3 MY III 10 22.5 32 NLF 0.0179 0.0174 1.03 131d N - Brightstock HaakeRV3 MY III 10 22.5 45.2 NLF 0.0213 0.0231 0.92 131e N- Brightstock HaakeRV3 MY III 10 22.5 64 NLF 0.0318 0.0306 1.04 131f N - Brightstock HaakeRV3 MY III 10 22.5 90.5 NLF 0.0419 0.0393 1.06 132a N- Brightstock HaakeRV3 MYIII 17 22.5 16 NLF 0.0118 0.0112 1.05 132b N - Brightstock HaakeRV3 MYIII 17 22.5 22.6 NLF 0.0153 0.0147 1.04

L__ 132c N- Brightstock Ha~~~y~_ MVIII 17 22.5 32 NLF 0.0203 0.0203-- 1.00 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle {wt 0/o) tC) (rpm) Flow (N-m) (N-m) torque 132d N - Brightstock HaakeRV3 MYIII 17 22.5 45.2 NLF 0.0274 0.0262 1.05 132e N - Brightstock HaakeRV3 MVIII 17 22.5 64 NLF 0.0353 0.0333 1.06 132f N - Brightstock HaakeRV3 MYIII 17 22.5 90.5 SLF 0.0438 0.0059 7.44 133a N - Brightstock HaakeRV3 MYIII 10 50 90.5 NLF 0.0045 0.0049 0.92 133b N- Brightstock HaakeRV3 MY III 10 50 128 NLF 0.0062 0.0061 1.02 133c N - Brightstock HaakeRV3 MYIII 10 50 181 NLF 0.0083 0.0081 1.02 133d N - Brightstock HaakeRV3 MY III 10 50 256 NLF 0.0110 0.0109. 1.01 ...... V'l 0 133e N - Brightstock HaakeRV3 MY III 10 50 362 NLF 0.0148 0.0146 1.02

133f N - Brightstock HaakeRV3 MYIII 10 50 512 NLF 0.0194 0.0190 1.02 I

133g N - Brightstock HaakeRV3 MY III 10 50 724 NLF 0.0272 0.0256 1.06 i 134a N - Brightstock HaakeRV3 MY III 17 50 90.5 NLF 0.0051 0.0049 1.05 134b N- Brightstock HaakeRV3 MVIII 17 50 128 NLF 0.0067 0.0064 1.05 134c N - Brightstock HaakeRV3 MY III 17 50 181 NLF 0.0088 0.0086 1.03 134d N - Brightstock HaakeRV3 MYIII 17 50 256 NLF 0.0115 0.0113 1.02 134e N- Brightstock HaakeRV3 MYIII 17 50 362 NLF 0.0150 0.0147 1.02 ----~~------Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt o/o) \C) (rpm) Flow (N-m) (N-m) torque 134f N - Brightstock HaakeRV3 MVIII 17 50 512 NLF 0.0198 0.0196 1.01 134g N- Brightstock HaakeRV3 MVIII 17 50 724 SLF 0.0287 0.0044 6.50 135a Bitumen HaakeRV3 MVIII 30 50 16 NLF 0.0304 0.0294 1.03 135b Bitumen HaakeRV3 MVIII 30 50 22.6 NLF 0.0385 0.0341 1.13

135c Bitumen HaakeRV3 MVIII 30 50 32 SLF 0.0453 0.0006 71.08 I VI"""""' 136 N - Brightstock HaakeRV3 MVIII 17 22.5 64 NLF 0.0419 0.0338 1.24 """""' 137a N- Brightstock RS 150 Z31 Ti 10 22.5 90.5 NLF 0.0449 0.0391 1.15 137b N- Brightstock RS 150 Z31 Ti 10 22.5 128 NLF 0.0521 0.0478 1.09 137c N - Brightstock RS 150 Z31 Ti 10 22.5 181 SLF 0.0639 0.0068 9.34 137d N - Brightstock RS 150 Z31 Ti 10 22.5 256 SLF 0.0035 0.0059 0.59 137e N - Brightstock RS 150 Z31 Ti 10 22.5 362 NLF 0.0058 0.0627 0.09 138 Shell flex 810 HaakeRV3 MVIII 30 30 362 NLF 0.0305 0.0246 1.24 139 N - Brightstock RS 150 Z31 Ti 17 22.5 90.5 SLF 0.0422 0.0032 13.04 140 Shell flex 810 HaakeRV3 MVIII 30 30 512 NLF 0.0368 0.0303 1.21 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt 0/o) (»C) (rpm) Flow (N-m) (N-m) torque 141 N - Brightstock RS 150 Z31 Ti 10 22.5 181 SLF 0.0642 0.0200 3.20 142a N - Brightstock HaakeRV3 MVIII 25 50 22.6 NLF 0.0018 0.0018 1.00 142b N - Brightstock HaakeRV3 MVIII 25 50 32 NLF 0.0027 0.0027 1.02 142c N- Brightstock HaakeRV3 MVIII 25 50 45.2 NLF 0.0034 0.0032 1.06 142d N - Brightstock HaakeRV3 MVIII 25 50 64 NLF 0.0044 0.0044 1.01 142e N - Brightstock HaakeRV3 MVIII 25 50 90.5 NLF 0.0058 0.0055 1.04 t--1 142f N - Brightstock HaakeRV3 MVIII 25 50 128 NLF 0.0076 0.0073 1.04 Vl N 142g N - Brightstock HaakeRV3 MVIII 25 50 181 NLF 0.0098 0.0093 1.05 142h N- Brightstock HaakeRV3 MVIII 25 50 256 NLF 0.0127 0.0118 1.07 142i N - Brightstock HaakeRV3 MVIII 25 50 362 NLF 0.0157 0.0139 1.13 142j N - Brightstock HaakeRV3 MVIII 25 50 512 SLF 0.0161 0.0012 13.16 143 Shell flex 810 RS 150 Z31 Ti 30 0 45.2 NLF 0.0539 0.0500 1.08 144a N - Brightstock HaakeRV3 MVIII 35 50 22.6 NLF 0.0029 0.0027 ~--- 1.07 Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to final Run No. Oil Device Spindle (wt o/o) tC) (rpm) Flow (N-m) (N-m) torque 144b N - Brightstock Haake RV3 MY III 35 50 32 NLF 0.0034 0.0033 1.04 144c N - Brightstock HaakeRV3 MYIII 35 50 45.2 NLF 0.0044 0.0043 1.02 144d N - Brightstock HaakeRV3 MYIII 35 50 64 NLF 0.0056 0.0055 1.03 144e N - Brightstock HaakeRV3 MYIII 35 50 90.5 NLF 0.0073 0.0071 1.02 144f N - Brightstock HaakeRV3 MY III 35 50 128 NLF 0.0091 0.0088 1.03 144g N - Brightstock HaakeRV3 MYIII 35 50 181 NLF 0.0114 0.0098 1.16 144h N - Brightstock HaakeRV3 MY III 35 50 256 NLF 0.0126 0.0107 1.18

I--' 144i N- Brightstock HaakeRV3 MY III 35 50 362 SLF 0.0125 0.0019 6.74 Ul w 145 Shell flex 81 0 HaakeRV3 MVIII 30 30 512 NLF 0.0418 0.0345 1.21 146 Shell flex 810 RS 150 Z31 Ti 30 30 724 SLF 0.0560 0.0044 12.62 147 N- Briglltstock HaakeRV3 MYIII 17 50 724 SLF 0.0294 0.0048 6.12 148 Shell flex 81 0 RS 150 Z31 Ti 30 0 64 SLF 0.0591 0.0038 15.73 149 N- Brightstock HaakeRV3 MYIII 30 30 90.5 SLF 0.0331 0.0020 16.88 150 N - Brightstock HaakeRV3 MY III 17 50 512 NLF 0.0228 0.0194 1.18 - -- Table B.l Couette flow experimental results for N-Brightstock, Shellflex 810 and Bitumen emulsions (cont'd)

Ratio Emulsion of Water Spindle Type of Initial Final initial Shearing Measuring concentration Temperature speed Lubricating torque torque to fmal Run No. Oil Device Spindle (wt o/o) \C) (rpm) Flow (N-m) (N-m) torque 151 N - Briglttstock HaakeRV3 MYIII 30 40 181 SLF 0.0227 0.0012 18.56 152 N - Brightstock HaakeRV3 MVIII 30 40 362 SLF 0.0328 0.0015 22.33 153 N- Brightstock HaakeRV3 MYIII 30 50 362 SLF 0.0466 0.0025 19.00 154 N - Brightstock RS 150 Z31 Ti 10 22.5 256 SLF 0.0785 0.0048 16.20 155 N - Brightstock HaakeRV3 MVIII 30 50 512 ... 90.5 156 N- Brightstock RS 150 Z31 Ti 10 22.5 128 NLF 0.0538 0.0450 1.19 157 N - Brightstock HaakeRV3 MYIII 30 22.5 45.2 ... 16 ~ Vl 158a Bitumen HaakeRV3 MY III 30 50 4 NLF 0.0039 0.0039 1.00 ~ 158b Bitumen HaakeRV3 MY III 30 50 8 NLF 0.0069 0.0069 1.00 158c Bitumen HaakeRV3 MYIII 30 50 16 NLF 0.0115 0.0113 1.02 158d Bitumen HaakeRV3 MY III 30 50 22.6 NLF 0.0147 0.0145 1.02 158e Bitumen HaakeRV3 MYIII 30 50 32 NLF 0.0188 0.0184 1.02 158f Bitumen HaakeRV3 MVIII 30 50 45.2 SLF 0.0238 0.0025 9.70 159 N- Brightstock HaakeRV3 MVIII 35 30 90.5 SLF 0.0277 0.0020 14.13 160 N - Brightstock HaakeRV3 MYIII 25 22.5 64 SLF 0.0377 0.0027 14.00 161 N- Brightstock HaakeRV3 MY III 30 30 90.5 SLF 0.0279 L. __ 0.0098 2.85 Table B.2 Reproducibility of Couette flow experimental results obtained with 30 wt% water-in-N-Brightstock oil emulsion

Ratio of Spindle Type of Initial Final initial to Shearing Temperature speed Lubricating torque torque fmal Run No. Device lC) (rpm) Flow (Nm) (Nm) torque 82b HaakeRV3 22.5 22.6 NLF 0.0221 0.0215 1.03 89 HaakeRV3 22.5 22.6 NLF 0.0201 0.0174 1.15 82c HaakeRV3 22.5 32 SLF 0.0277 0.0017 16.14 83 HaakeRV3 22.5 32 SLF 0.0247 0.0025 10.10 75d HaakeRV3 30 64 NLF 0.0260 0.0245 1.06 77 HaakeRV3 30 64 MLF 0.0314 0.0032 9.85 119 HaakeRV3 30 64 MLF 0.0240 0.0032 7.54 75f HaakeRV3 30 90.5 SLF 0.0319 0.0034 9.29 76 HaakeRV3 30 90.5 SLF 0.0389 0.0018 21.15 112 HaakeRV3 30 90.5 SLF 0.0299 0.0034 ·8.71 113 HaakeRV3 30 90.5 SLF 0.0319 0.0017 18.57 116 HaakeRV3 30 90.5 SLF 0.0307 0.0052 12.52 149 HaakeRV3 30 90.5 SLF 0.0331 0.0020 16.88 84f HaakeRV3 40 128 NLF 0.0191 0.0186 1.03 85 HaakeRV3 40 128 MLF 0.0181 0.0130 1.40 90 HaakeRV3 40 128 MLF 0.0194 0.0096 2.03 84g HaakeRV3 40 181 SLF 0.0243 0.0039 6.19 86 HaakeRV3 40 181 SLF 0.0260 0.0016 16.56 151 HaakeRV3 40 181 SLF 0.0227 0.0012 18.56

*Run numbers e.g. 64a, 82b indicate that the Stepwise spindle speed procedure was used. Abbreviations NLF represents Non-Lubricating Flow, MLF represent Marginal Lubricating Flow and SLF represents Self-Lubricating Flow.

155 Table B.2 Reproducibility of Couette flow experimental results obtained with 30 wt% water-in-N-Brightstock oil emulsion (cont'd)

Ratio of Spindle Type of Initial Final initial to Shearing Temperature speed Lubricating torque torque fmal Run No. Device c»c) (rpm) Flow (Nm) (Nm) torque 91f HaakeRV3 50 256 NLF 0.0137 0.0098 1.40 95 HaakeRV3 50 256 NLF 0.0196 0.0137 1.43 108 HaakeRV3 50 256 NLF 0.0186 0.0137 1.36 106 HaakeRV3 50 256 NLF 0.0178 0.0135 1.32 91g HaakeRV3 50 362 SLF 0.0137 0.0098 1.40 96 HaakeRV3 50 362 SLF 0.0216 0.0042 5.18 111 Haake RS150 50 362 SLF 0.0130 0.0026 5.05 117 HaakeRV3 50 362 SLF 0.0218 0.0029 7.42 107 HaakeRV3 50 362 SLF 0.0221 0.0026 ·8.33 122 HaakeRV3 50 362 SLF 0.0201 0.0020 10.25 97 HaakeRV3 50 512 SLF 0.0299 0.0049 6.10 110 HaakeRV3 50 512 SLF 0.0235 0.0015 15.97

156 Table B.3 Couette flow experimental results for 30 wt% water-in-N-Brightstock oil emulsion

Type of Spindle Initial Final Ratio of Lubricating speed Temperature Torque Torque initial to Run No. flow (rpm) ('C) (N_m) (Nm) rmal torque 104 NLF 8 15 0.0215 0.0184 1.17 99 NLF 16 15 0.0327 0.0253 1.29 100 SLF 22.6 15 0.0468 0.0053 8.81 101 SLF 32 15 0.0468 0.0059 8.00 103 SLF 45.2 15 0.0444 0.0028 15.73 82a NLF 16 22.5 0.0172 0.0171 1.01 82b NLF 22.6 22.5 0.0221 0.0215 1.03 83 SLF 32 22.5 0.0247 0.0025 10.10 68 SLF 45.2 22.5 0.0323 0.0025 13.20. 67 SLF 64 22.5 0.0385 0.0013 30.19 75a NLF 16 30 0.0093 0.0091 1.03 75b NLF 22.6 30 0.0123 0.0118 1.04 75c NLF 32 30 0.0159 0.0154 1.03 75d NLF 45.2 30 0.0206 0.0197 1.04 77 MLF 64 30 0.0314 0.0032 9.85 76 SLF 90.5 30 0.0389 0.0018 21.15 102 SLF 128 30 0.0266 0.0022 11.87 105 SLF 181 30 0.0270 0.0025 10.75

157 Table B.3 Couette flow experimental results for 30 wt% water-in-N-Brightstock oil emulsion ( cont' d)

Type of Spindle Initial Final Ratio of Lubricating speed Temperature Torque Torque initial to Run No. flow (rpm) \C) (Nm) ~m) rmal torque 84a NLF 22.6 40 0.0048 0.0047 1.02 84b NLF 32 40 0.0061 0.0064 0.96 84c NLF 45.2 40 0.0085 0.0086 0.99 84d NLF 64 40 0.0110 0.0111 1.00 84e NLF 90.5 40 0.0146 0.0147 0.99 90 MLF 128 40 0.0194 0.0096 2.03 86 SLF 181 40 0.0260 0.0016 16.56 87 SLF 256 40 0.0254 0.0025 10.38 152 SLF 362 40 0.0328 0.0015 22.33 88 SLF 512 40 0.0270 0.0022 12.22 91a NLF 45.2 50 0.0044 0.0045 0.97 91b NLF 64 50 0.0056 0.0054 1.05 92 NLF 90.5 50 0.0078 0.0105 0.74 93 NLF 128 50 0.0105 0.0091 1.16 94 NLF 181 50 0.0127 0.0118 1.08 95 NLF 256 50 0.0196 0.0137 1.43 96 SLF 362 50 0.0216 0.0042 5.18 97 SLF 512 50 0.0299 0.0049 6.10

158 Table B.4 Couette flow experimental data for Bitumen froth

Ratio of Type of initial to Temperature Spindle speed Lubricating Initial torque Final torque rmal ~C) (rpm) flow (Nm) (Nm) torque 35 11.3 NLF 0.0270 0.0247 1.09 35 16 SLF 0.0309 0.0002 126.00 35 22.6 SLF 0.0294 0.0002 120.00 35 32 MLF 0.0311 0.0093 3.34 35 45.2 SLF 0.0348 0.0002 142.00 40 11.3 NLF 0.0135 0.0127 1.06 40 16 SLF 0.0140 0.0002 57.00 40 22.6 SLF 0.0191 0.0002 78.00" 40 32 SLF 0.0247 0.0002 101.00 40 45.2 SLF 0.0306 0.0002 125.00 40 64 SLF 0.0348 0.0005 71.00 45 32 NLF 0.0255 0.0216 1.18 45 45.2 SLF 0.0333 0.0005 68.00 45 64 SLF 0.0274 0.0010 28.00 45 128 SLF 0.0417 0.0012 34.00 50 128 SLF 0.0336 0.0002 137.00 50 181 SLF 0.0451 0.0005 92.00 60 362 SLF 0.0179 0.0019 9.61 60 512 SLF 0.0103 0.0020 5.25 60 724 SLF 0.0113 0.0020 5.75

159 T a bl e B .5 Effiect o f water concentration on se lf.l- u b ncatmg . flow

Temperature Critical spindle tC) Water Concentration (wt%) speed (rpm) 22.5 10 181 22.5 17 90.5 22.5 25 64 22.5 30 32 22.5 35 32 30 25 128 30 30 64 30 35 64 50 17 724 50 25 512 50 30 362 50 35 362

160 Table B.6 Effect of initial torque reduction time on self-lubricating flow of30 wt% water-in-N-Brightstock emulsion

Temperature Spindle speed Time \C) (rpm) (sec) 15 22.6 2183 15 32 837 15 45.2 327.7 22.5 32 1900 22.5 45.2 2050 22.5 64 400 30 90.5 100 30 128 146.4 30 181 145.9 40 181 155 40 256 50 40 362 100 40 512 40 50 362 125 50 512 25

161 Table B.7 Couette flow data for 30 wt% N-Brightstock emulsion sheared with Z31 spindle (Shear rate, temperature as independent variables)

Spindle Temperature Spindle speed speed Shear rate Type of Run No. \C) (rpm) (radls) ( s·t) Lubricatin2 flow 104 15 8 0.84 3.54 NLF 99 15 16 1.68 7.08 NLF 100 15 22.6 2.37 10.00 SLF 101 15 32 3.35 14.16 SLF 103 15 45.2 4.73 20.00 SLF 82a 22.5 16 1.68 7.08 NLF 82b 22.5 22.6 2.37 10.00 NLF 83 22.5 32 3.35 14.16 SLF 68 22.5 45.2 4.73 20.00 SLF 67 22.5 64 6.70 28.32 SLF 75a 30 16 1.68 7.08 NLF 75b 30 22.6 2.37 10.00 NLF 75c 30 32 3.35 14.16 NLF 75d 30 45.2 4.73 20.00 NLF 77 30 64 6.70 28.32 MLF 76 30 90.5 9.48 40.05 SLF 102 30 128 13.40 56.64 SLF 105 30 181 18.95 80.10 SLF

162 Table B.7 Couette flow data for 30 wt% N-Brightstock emulsion sheared with Z31 spindle (Shear rate, temperature as independent variables) [ cont' d]

Spindle Temperature Spindle speed speed Shear rate Type of Run No. ~C) (rpm) (rad/s) ( s·•) Lubricating flow 84a 40 22.6 2.37 10.00 NLF 84b 40 32 3.35 14.16 NLF 84c 40 45.2 4.73 20.00 NLF 84d 40 64 6.70 28.32 NLF 84e 40 90.5 9.48 40.05 NLF 90 40 128 13.40 56.64 MLF 86 40 181 18.95 80.10 SLF 87 40 256 26.81 113.29 SLF 88 40 512 53.62 226.58 SLF 91a 50 45.2 4.73 20.00 NLF 9lb 50 64 6.70 28.32 NLF 92 50 90.5 9.48 40.05 NLF 93 50 128 13.40 56.64 NLF 94 50 181 18.95 80.10 NLF 95 50 256 26.81 113.29 NLF 96 50 362 37.91 160.20 SLF 97 50 512 53.62 226.58 SLF

163 Table B.8 Couette flow data for 30 wt% N-Brightstock emulsion sheared with Z38 spindle (Shear rate , temperature as independent variables)

Spindle Type of Temperature Spindle speed speed Shear rate Lubricating Run No. \C) (rpm) (rad/s) ( s-t) flow 115a 22.5 8 0.84 7.28 NLF 115b 22.5 16 1.68 14.56 NLF 115c 22.5 22.6 2.37 20.56 NLF 115d 22.5 32 3.35 29.11 SLF 115e 22.5 45.2 4.73 41.12 SLF 115f 22.5 64 6.70 58.22 SLF 120a 40 8 0.84 7.28 NLF 120b 40 16 1.68 14.56 NLF 120c 40 22.6 2.37 20.56 NLF 120d 40 32 3.35 29.11 NLF 120e 40 45.2 4.73 41.12 NLF 120f 40 64 6.70 58.22 NLF 120 40 90.5 9.48 82.33 NLF 120h 40 128 13.40 116.45 NLF 120i 40 181 18.95 164.66 NLF 120j 40 256 26.81 232.89 SLF 120k 40 362 37.91 329.32 SLF

164 Table B.9 Couette flow data for 30 wt% N-Brightstock emulsion sheared with Z41 spindle (Shear rate, temperature as independent variables)

Spindle Temperature Spindle speed speed Shear rate Type of Run No. \C) (rpm) (radls) ( s·•) Lubricating flow 114a 22.5 8 0.84 19.40 NLF 114b 22.5 16 1.68 38.79 NLF 114c 22.5 22.6 2.37 54.79 NLF 123a 40 8 0.84 19.40 NLF 123b 40 16 1.68 38.79 NLF 123c 40 22.6 2.37 54.79 NLF 123d 40 32 3.35 77.58 NLF 123e 40 45.2 4.73 109.59 NLF 123f 40 64 6.70 155.17 NLF 123g 40 90.5 9.48 219.41 NLF 123h 40 128 13.40 310.33 NLF

165 Table B. tO Effect of oil viscosity on self-lubricating flow

Temperature Viscosity Critical spindle Oil c-c> (Pa.s) speed (rpm)

N - Brightstock 15 32.51 22.6

22.5 14.71 32

50 1.04 362

Shellflex 81 0 0 13.63 64

30 0.98 724

Bitumen 50 25.63 32

166 60

...... 50 ......

-40 0 0 -! 1U30:l ~ a.CD E CD ll ll t- 20 ..... • •

+Non - Lubricating flow 10 - 1::,. Self- Lubricating flow

0 I 0 100 200 300 400 500 600 700 800 Spindle speed (rpm)

Figure B.l Flow map for 10 wt% water-in-N-Brightstock oil emulsion (Temperature and spindle speed as independent variables).

167 60

...... A 50 T u ""' ""' """ """ ""'

.-.40 (.) 0 ~e ~ fti30.. c.CD E CD .... ll 1-20

• Non - Lubricating flow

1::. Self - Lubricating flow 10 f---

0 r 0 100 200 300 400 500 600 700 800 Spindle speed (rpm)

Figure B.2 Flow map for 17 wt% water-in-N-Brightstock oil emulsion {Temperature and spindle speed as independent variables).

168 00

...... A 50 ...... L..l

...... A ...... L..l

+l'b'l - Lubricating flON 8 Margnal -Lubricating flON 10 - fl Self- Lubricating flON

0 I I I 0 100 200 300 400 500 000 Spindle speed (rpn)

Figure B.3 Flow map for 25 wt% water-in-N-Brightstock oil emulsion {Temperature and spindle speed as independent variables).

169 00

...... A 50 ...... ~

...... A A ...... u ~

·~ ~ ~

• ~- L.ubicating flON 10 ~ A Self -I.J.iJricating flo,v

0 I I I 0 50 100 150 200 250 300 350 400 Spindle speed (rpm)

Figure B.4 Flow map for 35 wt% water-in-N-Brightstock oil emulsion (Temperature and spindle speed as independent variables).

170 APPENDIXC Derivation of the Viscous Heating Model

The details of the derivation of the viscous heating model presented in Chapter 4 is presented thus:

Assumptions

1. Adiabatic system (oil acts as a perfect insulator) 2. Work done by the fluid in exerting effect on rotating spindle is completely

converted into heat required to raise the fluid temperature by ~roc.

Power (rate of work)= Torque x angular velocity

= Tm

Energy= Power x Time = T mt

Where tis duration of shear (s)

Energy required to raise temperature of the fluid

= mCpfluid~T

Thus, from the second assumption,

(C.l)

and ~T = Tmt (C.2) mCpfluid

From equation (C.2),

T = _m_C_p;._flu_id~_T_ (C.3) aJt

171 Equation previously derived for Newtonian fluid in Couette flow in a concentric cylinder viscometer is given by,

(2.1)

Equating (C.3) and (C.4)

(C.4)

fl.T = 4njJLtoi (C.5) u2 -~ Jmcp flaM

172 Appendix D Derivation of the Models for Self-Lubricating Flow in a Couette Cell

Model I has already been presented in Chapter 5. The detailed derivation of Model II is presented in this Appendix. Model II assumes shearing throughout the Couette Cell gap. A schematic illustration of Model II is shown in Figure D.l. The boundary conditions are stated below:

Boundary conditions

At r = R~, VeA = Rtro At r = R2, veA = vea; trtlA = trt)a At r = R3, vea = vee; trt)B = trtlc At r = RJ, voc=O

The fluids will be assumed to be Newtonian and incompressible and at steady state and laminar flow conditions.

173 Model II - Shearing Throughout Couette Cell Gap

~

+ Spindle

Figure D.l Schematic illustration of Model II (Shearing in water and emulsion layers). Emulsion layer- A and C; Water layer B

174 Based on the equations for the Couette flow of a single phase fluid in a viscometer; the shear stress and velocity is given by the following equations (Appendix E) cl 'rro = -2 (E.4) r

(E.8)

The velocity profiles for Fluids A , B and C can thus be expressed in terms of Equation E.8.

For fluid A, this is

cl v8 =--+C2r Rt

For fluid B, this is

DI v8 =--+D2r R2

For fluid C, this is

Et v8 =--+E2r R3

Applying the boundary conditions to Equations 0.1 to 0.3;

(0.4)

(0.5)

(0.6)

175 (0.7)

(0.8)

(0.9)

From equations (0.6) and (0.8),

(0.10)

(0.11)

Thus, (0.12)

From equation (0.4),

(0.13)

from equation (0.9),

(0.14)

Applying equations (0.11) and (0.12) to equation (0.7) gives

(0.15)

176 Applying equations (D.lO) and (D.12) to equation (D.5) gives

(D.16)

where for simplicity, Let

K=~~~~~-~~~~~-~~~~~-~~~~~ + J.l AJ.l nR[ ~ Ri + J.l AJ.lcR[ RiR;

Dt and Et are also given by equation (D.16)

And thus from (D.13),

C = Kco + J.laJ.lc~RiR;co (D.l7) 2 K

Substituting C 1 from equation (D.16) into equation (D.15)

(D.l8)

Substituting E1 = C~, and C1 from equation (D.l6) into equation (D.14) gives

(D.l9)

177 Substituting JlA = J.lc and the constants given by equations (D.16) to (D.l8) into equations (D.l ), (D.2) and (D.3) gives

(D.20)

(D.21)

(D.22)

Shear stress is given by equation (D.l).

Thus, substituting Ct into equation (E.4) gives

(D.23)

Torque exerted on the spindle

or

2 T = - 4;q.J~JJBR1 ~ RJ RiLm (D.24) K where

K=~~~~~-~~~~~-~~~~-~~~~~ II II R2R2R2 + 112 R2 1)2R2 + rArB I 2 3 rA 1.£~ 4

178 APPENDIX E Derivation of the Equations for Single Phase Couette Flow in A Viscometer

The equations governing single phase Couette flow in a concentric cylinder viscometer will be reviewed.

A concentric cylinder system is illustrated in Figure E.l. The fluid is placed in a stationary cup of radius R2 and is sheared across the gap between the spindle and cup by a spindle of radius R1 (also known as a bob or rotor) rotating at an angular velocity ro.

A large gap between the spindle and the cup will be assumed. Steady state and laminar flow will be assumed. The fluid will also be assumed to be incompressible and Newtonian.

Figure E.l Couette flow of a single phase, incompressible Newtonian fluid in a concentric cylinder viscometer.

179 The equation of continuity in cylindrical coordinates taken from Table 3.4-1 (Bird et al., 1960) is

(E.1)

The assumptions of steady state and a velocity component only in the r direction reduces this equation to zero.

Also, the equation of motion in cylindrical coordinates in terms of 1: taken from Table 3.4-3 (Bird et al., 1960) is r-component

9-component

P(Ovo + v Ovo + Vo Ovo + vrvo + v Ovo) = _.!_ Bp ot r ar r ao r z oz r a(}

1 a ( 2 ) 1 00 - -- r 'ro +---+-ar a,(k) + pgo ( r2 or r ao oz

z-component

180 (E.2)

Since motion is in the 9 direction, solving the equation of motion in the 9 direction will yield the velocity profile.

The assumptions simplify the 9-component to

(E.3)

Solving Equation E.3 gives

(E.4)

where Ct is a constant of integration.

From Table 3.4-6 (Bird et al., 1960),

(E.S)

since there is no velocity component in the r direction, Equation E.S becomes

181 (E.6)

Substituting Equation E.6 in Equation E.4;

Ct = -p[r!_(vo )] (E.7) r2 or r which gives

(E.8)

where C1 and C2 are constants of integration

Solving Equation E.8 with boundary conditions r =R~, vo = Rtro and r = R2, vo = 0 yields

Equations E.4 and E.8 thus become

(E.9)

which is an expression for the shear stress.

182 (E.lO)

which represents the velocity profile as a function of radial position.

The shear rate is thus given by

(E.ll)

The Torque T required to turn the spindle is given by

T = rr8 (r = R ) * 21rR L * R 1 1 1 (E.l2) where L is the length of the spindle which eventually yields

(E.l3)

And Spindle angular velocity, 2tcn OJ=- (E.14) 60 Spindle speed, n has units of revolutions per minute (rpm)

183 The equation for a Bingham fluid in a concentric cylinder geometry is given by Shook and Roco (1991) as

(E.15)

The equations E.1 0 and E.11 for the velocity and shear rate profiles will be illustrated using the dimensions of the Haake RS 150 rheometer which was used in this research. The dimensions are given below: Radius Rt Small spindle radius = 15.725 mm Medium spindle radius= 19.01 mm Large spindle radius= 20.71 mm

Cup radius R2 = 21.667 mm Gap width = R2 - Rt Small spindle gap= 5.942 mm Medium spindle gap = 2.657 mm Large spindle gap = 0.957 mm

An arbitrary spindle speed of 50 rpm (5.24 rad/sec) was used in the calculations. The velocity and shear rate profiles are illustrated in Figures E.2 and E.3 respectively. Velocity and shear rate are normalized with the maximum velocity and maximum shear rate at the spindle respectively. Radial position is normalized with the gap width. It should be noted that radial position as shown increases from the spindle to the cup.

184 1 ~------~

-+-Small spindle (15.725 mm radius) -a-Medium spindle (19.01 mm radius) -e-Large spindle (20.71 mm radius)

-~ ·u 0.3+------~~~------; 0 ~ 0.2 +------~~~------~

0+-----~----~----~----~----~----~--~~--~----~----~ 0 0.1 0.2 0.3 0.4 0.5 . 0.6 0.7 0.8 0.9 1 (r • R1) I (R2· R1)

Figure E.l Couette Flow Velocity profile in a concentric cylinder viscometer as a function of spindle size and radial position progressing from spindle towards cup.

Radius R1 is spindle radius; R2 is cup radius (21.667 mm). Spindle speed = 50 rpm.

185 1

0.9 -.! l! 0.8 ... C'G .! 0.7 tn E::s 0.6 E ·= 0.5 :E._...... 0.4 -+- Small spindle (15.725 rrm radius) -.! l! 0.3 +------1 -a- Meditm spindle (19.01 rrm radius) ... -+-Large spindle (20.71 rrm radius) C'G Cl) .c 0.2 tn -0.1

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 (r- Rt) I (R2 • Rt)

Figure E.3 Couette Flow Shear rate profile in a concentric cylinder viscometer as a function of spindle size and radial position progressing from spindle towards cup.

Radius Rt is spindle radius; R2 is cup radius (21.667 mm). Spindle speed = 50 rpm.

186 It can be observed from Figure E.2 that the velocity is a maximum at the spindle and decreases to zero at the cup. It can be observed from Figure E.2 that the velocity profile for the large spindle appears almost linear and thus implies an approximately constant shear rate. But the shear rate profile for the large spindle in Figure E.3 shows a variation of about 10% from the spindle to the cup. It would thus not be appropriate to assume a constant shear rate profile for the large spindle. Calculations show that a spindle gap width of less than 0.067 mm is required to give a less than 1% variation in shear rate in this case.

187